JP3821397B2 - Image cropping device - Google Patents

Description

【００４１】
上記の条件設定領域の指定及び該指定された領域に基づく切り抜き条件の設定が終了すると（Ｓ３）、次に実際に画像の切り抜きを行う領域（マスク領域）の指定を、前記入力手段６によって行う（Ｓ４：切り抜き領域指定手段）。
前記切り抜き処理領域の指定においては、条件設定領域の指定と同様に、オペレータがＣＲＴ５上の原画像を見ながら、主要画像と背景部分との境界部分を含むような領域を、ポインティングデバイスによる座標指定によって指定し、設定された切り抜き処理領域は、ＣＲＴ５上の原画像に重ねて表示される。
【００４２】
処理領域の指定においては、なるべく領域を狭くすることが自動輪郭抽出処理の負担を少なくすることになるから、図４に示すように境界部分を複数領域に分けて指定することが好ましく、また、輪郭部分を洩れなく処理領域に含めるために、指定される処理領域間における重なりを許容し、最終的には指定された複数の処理領域を包含する領域を処理領域とすると良い。
【００４３】
また、指定済みの条件設定領域と、前記処理領域の指定とが一部重なったり、或いは、条件設定領域を含む形での処理領域の指定が許容されるようにすると良い。
更に、切り抜きを行う領域の指定においても、図４に示すような四角な領域に限定されず、任意の形状による領域指定を行えるようにしても良い。
【００４４】
また、切り抜きたい主要画像が小さい場合には、主要画像を包含するように切り抜き処理を行う領域を指定しても良い。
切り抜き処理の領域指定が終了すると、前記設定された切り抜き条件に従って前記切り抜き処理領域内で主要画像の輪郭線を自動抽出し（Ｓ５：画像切り抜き手段）、かかる自動抽出処理が正規に行われた場合には（Ｓ６）、該抽出結果が確定し、以て、マスクの作成がなされる（Ｓ５）。
【００４５】
具体的には、条件設定領域内の画像データから画像データの閾値として切り抜き条件を設定した場合には、指定された処理領域内の画像データを前記閾値と比較し、主要画像に対応する画素と背景部分に対応する画素とに区分することで、主要画像の輪郭を抽出する。
即ち、上記実施例では、指定された複数或いは１つの条件設定領域内の画像データに基づいて複数の処理領域に共通の１つの切り抜き条件を設定し、複数の処理領域において一定の条件の下に切り抜き処理を実行させるようにしてある。
【００４６】
従って、複数の条件設定領域に基づいて切り抜き条件が設定された場合には、変化する背景部分を総合的に判断した適正な切り抜き条件の設定が可能であり、更に、複数の処理領域で共通的に１つの切り抜き条件を用いることで、各処理領域で統一的な切り抜き処理が行える。
ここで、前記抽出された輪郭線を、ＣＲＴ５上に原画像に重ねて表示し、オペレータに自動抽出結果が正しいか否かを判断させるようにすると良い。そして、自動抽出された輪郭に不備がある場合には、ポインティングデバイスによる手動の輪郭修正を加えたり、切り抜き条件の手動修正を加えることができるようにしたり、或いは、条件設定領域の指定をやり直すことができるようにすると良い。
【００４７】
上記図５のフローチャートに示す手順では、先に条件設定領域を指定し、該指定された条件設定領域（複数或いは単数）に含まれる画像データに基づいて決定される切り抜き条件に従って、その後に指定される切り抜き処理領域（複数或いは単数）で主要画像の輪郭抽出を行わせる構成としたが、図６のフローチャートに示すように、先に切り抜き処理領域を指定し、これらの領域での切り抜き処理の条件を設定するための条件設定領域を後から指定させる構成としても良い。
【００４８】
図６のフローチャートにおいて、まず、主要画像の輪郭部分を含む切り抜き処理領域（複数或いは単数）の指定（図４参照）を、前記図５のフローチャートの場合と同様にしてオペレータがポインティングデバイスを用いて行う（Ｓ11：切り抜き領域指定手段）。
主要画像の輪郭を全て網羅するようにして処理領域の指定が終了すると（Ｓ12）、続いて該指定された処理領域での切り抜き処理条件を設定するための条件設定領域の複数或いは単数指定を（図４参照）、前記図５のフローチャートの場合と同様にしてオペレータがポインティングデバイスを用いて行う（Ｓ13：条件設定領域指定手段）。
【００４９】
ここでも、処理領域に含まれるような条件設定領域の設定（処理領域内での背景部分の指定）、或いは、処理領域と一部重複するような条件設定領域の設定が許容されるようにすると良い。
条件設定領域の指定（複数領域或いは単数領域）が終了すると、かかる指定された条件設定領域の画像データ、即ち、主要画像の背景部分に対応する画像データに基づいて主要画像を切り抜くとき（輪郭抽出）に用いる切り抜き条件を設定する（Ｓ14：切り抜き条件設定手段）。
【００５０】
複数の条件設定領域が設定された場合には、前記同様に、指定順に従って領域に重み付けして切り抜き条件を設定させるようにしても良い。
また、上記図６のフローチャートに示される手順では、条件設定領域の指定時には既に処理領域が指定されているから、切り抜き条件の設定時においては条件設定領域と処理領域との相関が特定されていることになる。そこで、処理領域と条件設定領域とが重なる場合には、該重複領域の画像データは、主要画像の輪郭部分に近い背景部分であるから、被重複領域の画像データに対して前記重複領域の画像データに重み付けして、切り抜き条件の設定においては、前記重複領域の画像データが大きく影響するようにすると良い。
【００５１】
切り抜き条件の設定が終了すると、実際にかかる条件に従って、指定された各処理領域内で主要画像の輪郭線を抽出させ、マスク作成を行わせる（Ｓ15：画像切り抜き手段）。
尚、切り抜き処理領域を複数指定した後、或いは、切り抜き処理領域を指定する毎に、各処理領域毎に切り抜き条件を設定するための領域を特定して指示し、又は、条件設定領域を複数指定した後、或いは、条件設定領域を指定する毎に、各条件設定領域に基づく切り抜き条件で切り抜きを行わせる領域を特定して指示し、例えば処理領域Ａについては条件設定領域ａの画像データに基づいて切り抜き条件を設定させ、処理領域Ｂについては条件設定領域ｂの画像データに基づいて切り抜き条件を設定させるようにして、処理領域毎に異なる条件設定領域を対応させるようにしても良い。
【００５２】
かかる構成によれば、主要画像の背景部分が部位によって大きく変化する場合であっても、背景の変化に応じた複数の切り抜き条件の下で、対応する輪郭部分を精度良く抽出させることが可能となる。
更に、例えば処理領域Ａについては条件設定領域ａ，ｃの画像データに基づいて切り抜き条件を設定させ、処理領域Ｂについては条件設定領域ｂ，ｃの画像データに基づいて切り抜き条件を設定させなど、各処理領域毎に対応させる条件設定領域の組み合わせは異なるものの、一部重複する条件設定領域が存在する構成としても良い。
【００５３】
また、ＣＲＴ５上に処理領域，条件設定領域を表示するときに、各領域に識別符合を付して、処理領域と条件設定領域との対応関係の確認を容易にし、また、対応関係の指定を容易にすることが好ましい。
ところで、上記のようにして、オペレータが切り抜きの輪郭部分を含む領域を大まかに指定するときには、自動輪郭抽出処理の負担を軽減し、また、精度を向上させるために、主要画像の輪郭に沿ったなるべく狭い領域を簡易に指定できるようにすることが望まれる。更に、輪郭抽出がオペレータの領域指定の意図を反映して行われ、かつ、輪郭抽出の結果が領域指定に追従して容易に確認できるようにすることが望まれる。
【００５４】
そこで、以下に前記輪郭部分の領域指定及び輪郭抽出のより好ましい実施例を説明する。尚、以下に説明する第２実施例は、既述の第１実施例における図３に示されるハードウェア構成を参照しつつ説明する。
図７のフローチャートは、第２実施例におけるマスク作成の基本手順を示すものであり、まず、オペレータがＣＲＴ５（表示手段）上の原画像を見ながら入力手段６（ポインティングデバイス）を介して、主要画像の輪郭線上の１点を輪郭抽出を行わせる始点の座標位置として入力する（Ｓ21：輪郭座標指定手段）。続いて、同様に入力手段６によって前記始点とは異なる主要画像の輪郭線上の１点を、輪郭抽出を行わせる終点位置として座標入力する（Ｓ22：輪郭座標指定手段）。
【００５５】
かかる始点・終点位置の座標入力によって、図８に示すように、始点を中心とする所定半径範囲と、終点を中心とする所定半径範囲とを直線的に結ぶような切り抜き輪郭に沿った帯状領域が、輪郭抽出領域（切り抜き領域）として設定され、かかる帯状領域がＣＲＴ５上で原画像に重ねて線画表示される。そして、演算手段２によって前記抽出領域内で輪郭抽出が行われ、抽出された輪郭がやはりＣＲＴ５上で画像に重ねて表示される（Ｓ23：帯状領域設定手段，輪郭線抽出手段，輪郭線表示手段）。
【００５６】
即ち、前記第１実施例では、輪郭線を囲むように領域を指定したのに対し、第２実施例では、輪郭線上の主要点を指定することで、輪郭線を含む輪郭抽出領域が設定される構成であり、オペレータが主要画像の輪郭線をなぞるようにして、輪郭線上の主要点をマウス等のポインティングデバイスで指定することで、輪郭を含む狭い領域を簡便に指定できるものであり、然も、本実施例における帯状領域では、主要画像の輪郭線が帯状領域の幅方向の中央付近に位置する可能性が高いから、帯状領域からの輪郭抽出が容易に行える。
【００５７】
上記のような主要画像の輪郭上の主要点の座標位置の入力に伴う帯状領域（輪郭抽出領域）の設定により一連の輪郭点抽出が終了するまでは（Ｓ24）、入力手段２により座標位置が入力された点を順次終点位置として更新設定し、輪郭抽出を行わせる（輪郭線抽出手段）。
即ち、切り抜きたい画像の輪郭線上の点を最初に指定すると、それがまず始点として設定され、その後、輪郭線上の異なる点を指定すると、前記始点と終点とを結ぶ帯状領域が設定され、かかる帯状領域内で輪郭抽出が行われる。続いて、新たに輪郭線上の点が指定されると、前回に終点位置として指定した輪郭線上のポイントから新しい指定点（終点）にまで帯状領域が延設され、該延設された帯状領域内で輪郭抽出が行われ、一連の輪郭線を網羅する帯状領域の設定が終了するまで、輪郭線上の点の座標入力を繰り返す。
【００５８】
ここで、終点位置の更新設定があった場合には、前回までに求められている輪郭を一旦キャンセルし、全帯状領域内において新たに輪郭抽出を行わせるようにすると良い。これは、輪郭抽出において複数の輪郭点候補が検出され、これらの中から最終的に１つの輪郭点を選択するときに、終点位置の更新設定が、輪郭線の方向を示唆することになり、これが選択すべき輪郭候補点に変化を与えることになるためである。また、かかる終点位置の指定による輪郭抽出の方向付けにおいて、抽出された輪郭点がＣＲＴ５上に表示されるから、オペレータはかかる輪郭点の表示に基づいて抽出される輪郭線を所望の方向に導くように終点位置を意図的に指定することも可能となる。
【００５９】
一連の輪郭線が抽出されると、新たに輪郭抽出の始点を指定し（Ｓ25）、同様にして帯状領域の設定及び輪郭抽出を行わせ、全ての輪郭抽出が終了するまで上記のような処理を繰り返す（Ｓ26）。
そして、輪郭の抽出が全て終了すると、抽出された輪郭点をベクトル化（各輪郭点を連結）して、マスク画像を作成する（Ｓ27：画像切り抜き手段）。
【００６０】
次に、図９のフローチャートに従って、前記図７のフローチャートのＳ23における輪郭抽出の様子を詳細に説明する。
まず、前述のように、切り抜きを所望する画像の輪郭上で座標入力された主要点 (始点と終点) に基づいて、輪郭点の検出を行う帯状領域を設定する（Ｓ31：帯状領域設定手段）。
【００６１】
次に、原画像を光電変換して得られたカラー画像データ（Ｒ，Ｇ，Ｂ or Ｙ，Ｍ，Ｃ，Ｋ）のうちの帯状領域内の画像データを、公知の方法によりＨ（色相），Ｓ（彩度），Ｌ（明度）のデータに変換する（Ｓ32）。
そして、輪郭点の抽出に用いる複数の画像データの正規化定数を設定する（Ｓ33）。具体的には、画像データそれぞれの最大値と最小値との幅が同じになるように、例えば、画像データそれぞれの最大値Ｘ_max 及び最小値Ｘ_min を求め、Ｘ＝Ｘ／ (Ｘ_max −Ｘ_min ）× 100 によって正規化する (ここで、Ｘ＝Ｒ，Ｇ，Ｂ，Ｈ，Ｓ，Ｌ）。
【００６２】
続いて、帯状領域の前記始点と終点とを結ぶ中心線と交差する幅方向に画像データを始点側から走査し（図10参照）、当該走査線上（画像データ列上）で主要画像の輪郭線と交差する輪郭点の候補を算出する（Ｓ34：輪郭線抽出手段）。具体的には、画像データ毎（各色プレーン毎）に前記走査線（画像データ列）上で隣接する画素間でのデータの差分を求め、該差分が所定以上である画素位置として輪郭候補点を求める。
【００６３】
尚、輪郭点の抽出を、前記画像データＲ，Ｇ，Ｂ，Ｈ，Ｓ，Ｌを用いる構成に限定するものではなく、また、色プレーン毎に画素間のデータ変化を算出するのではなく、色プレーン間でのデータ偏差を求めて輪郭点を求める構成としても良い。
更に、前記第１実施例と同様に、帯状領域（切り抜き処理領域）の指定とは別に、主要画像の背景部分を含む領域（条件設定領域）を指定させ、該指定された領域に含まれる画像データに基づいて前記走査線上での輪郭点の算出に用いる閾値（切り抜き条件）を設定させるようにしても良い。
【００６４】
かかる輪郭点の算出を、終点位置が設定される毎に、始点から終点に向けて一定間隔（１画素単位或いは複数画素長さ）で行わせ、各走査線上で求められた輪郭候補点の連続性（連結性）を評価して、始点から終点に向けて各走査線上における輪郭候補点を選択する（Ｓ35）。そして、前記選択された輪郭候補点をベクトル化（各輪郭点を連結）して輪郭線を得て（Ｓ36）、該輪郭線をＣＲＴ５（表示手段）上に原画像に重ねて表示する（Ｓ37：輪郭線表示手段）。
【００６５】
オペレータは、指定した始点，終点位置に基づいて抽出された輪郭線を、前記ＣＲＴ５上の表示によって目視確認することができ、これによって、自動抽出された輪郭線の適否を判断できると共に、その後の輪郭抽出結果を予測することが可能となるから、表示された抽出結果を終点位置の更新設定の指標とするこができ、以て、抽出される輪郭線を終点位置の指定によって所望の方向に導くことが可能である。
【００６６】
即ち、表示された輪郭の抽出結果から、切り抜きを行いたい本来の輪郭線と異なる点をトレースしている場合や、誤った輪郭部分をトレースする可能性が高いことを判断できるから、この場合には、終点位置の新たな指定において前記誤ってトレースしている部分や誤ってトレースされる可能性の高い部分を帯状領域から除くようにしたり、或いは、終点位置の指定位置の間隔を変更したりして、所期の輪郭線がトレースされるように工夫することが可能となる。そして、終点位置の指定毎に始点から輪郭点の検出をやり直すから、連続性の評価が新たな終点位置の設定によって変化し、所期の輪郭線をトレースさせることができる。
【００６７】
次に、前記切り抜き処理制御を基本としたより好ましい態様を説明する。
まず、始点，終点位置の指定に基づいて設定される前記帯状領域の形状は、図８に示したように、円形の中心を前記設定された始点と終点との間で移動したときの円形外郭の軌跡形状として設定し、基本的には２つの円を直線状に結んだ形状とする。このような領域により、帯状領域の方向によらず、一定の幅 (円形の直径) に輪郭抽出領域が設定されるので輪郭線の抽出精度を一定に保持できる。ここで、該帯状領域の幅は、可変に設定されるのが好ましく、次のように自動的に設定することで、輪郭線の抽出精度を高めることができる。
【００６８】
例えば、帯幅設定方式としては、帯幅を表示された切り抜き対象画像の大きさに対して所定の比率 (画素数比) 、例えば１／20となるように設定する。これは、画像の読み込み解像度によって画像の像構造を表示する画素値の空間的変化が異なるため、帯幅が画像の切り抜き主体と一定の割合となるように設定する。例えば、同一画像を等倍で読み込んだ画像と、400 ％で読み込んだ画像において、同じ輪郭線を構成する隣接画素値の並びは大きく異なる。本実施例では、オペレータが指定する被切り抜き画像領域と画像サイズから切り抜きに適切な帯幅を選択する (１：20) 。なお画素比は、１：15〜１：25が好ましく、これより下方ではキズ等のノイズの影響を受けやすく推定輪郭線が迷走し、これより上方では適切な輪郭点を抽出しずらくなり、特に１：18〜１：22が好ましい。この場合、表示された切り抜き画像の大きさ (表示画素数) は当該画像の輪郭線を知らないと得られないので、例えば切り抜き画像に略外接する方形領域を設定し、該領域の大きさ (画素数) に基づいて帯幅を設定するような構成とすれば良い。
【００６９】
かかる実施例 (図９のＳ31のサブルーチン) を図11に示す。
図において、まず、切り抜き画像の表示サイズＳを読み込み（Ｓ41）、次いで、帯幅Ｗを１／20・Ｓ (画素数比) として設定し（Ｓ42）、該設定された帯幅Ｗで始点と終点とを結ぶ帯状領域を設定すると共に、該帯状領域をＣＲＴ５に表示する（Ｓ43）。
【００７０】
また、同じ大きさの画像でも局所的に輪郭線の変化率が大きくなる部分もあることを考慮して、手動でも帯幅Ｗを設定できるようにしておくのが良い。
ところで、帯状領域を直線形状のみに設定すると、曲率の大きい輪郭線に対して、始点，終点間を短い帯状領域で繋げて何回も設定する必要がある。そこで、オペレータの指示により帯状領域を湾曲化して設定することにより、１つの帯状領域の中に長い輪郭線を含ませることができ、設定回数を減らすことができる。具体的な湾曲化方法としては、例えばベジェー関数を用いることができる。
【００７１】
かかる実施例（図９のＳ31のサブルーチン）を図12に示す。
図において、まず、前記した方法等により帯幅Ｗを設定し（Ｓ51）、次いで、帯状領域を湾曲させるか否かのオペレータの指示を読み込み（Ｓ52）、湾曲しない場合は、そのまま直線状の帯状領域を表示し（Ｓ54）、湾曲させる場合は、前記ベジェー関数等を用いた湾曲処理を行うと共に（Ｓ53）、該湾曲処理された帯状領域をＣＲＴ５上に表示させる（Ｓ55）。そして、オペレータが湾曲処理された帯状領域の適正を判断し（Ｓ56）、ＯＫとなるまで湾曲処理を続け、ＯＫとなれば、該湾曲された帯状領域を最終設定する（Ｓ57）。
【００７２】
一方、同一の走査線上で輪郭候補点が複数検出された場合に、これらの中から真に所望する輪郭点を決定する方法として、帯状領域の幅方向（走査線方向）に予め重みを付けておくことが好ましい。
即ち、オペレータが帯状領域を設定する際、無意識に帯状領域の中心線を所望の切り抜き輪郭線に近づけて設定する確率が高いことを考慮して、帯状領域の幅方向の中心近傍の画像データに対する重みを大きく、端の方では重みを小さく設定し、前記複数の輪郭点候補がある場合は重みの大きい方を選択する。但し、始点と終点では、輪郭線が帯幅の中心近傍にある確率が高いが、中間部では中心から離れる可能性も高まるので、始点及び終点に近い部分のみ重み付けしたり、重み付けの度合いを始点或いは終点からの距離によって変化させるようにしても良い。
【００７３】
また、帯幅走査方向の濃度や色の変化が小さく前記各データの差分からは実質上輪郭点を判別することが不能であるような場合は、かかる判別不能部分について帯状領域の延設方向に従って図13に示すように外挿により輪郭線を決定する。また、決定された輪郭点が前回の走査線上で決定された輪郭点の近傍にあるか否かを判定し、近傍に無い場合は、輪郭点の決定が誤っていると判断して前記同様外挿により輪郭点を決定するか、或いは、輪郭点候補が同一走査線上で複数ある場合は、２番目の候補を前回決定された輪郭点と比較し、その近傍にあれば、１番目の候補に代えて２番目の候補に選択し直すようにしてもよい。
【００７４】
更に、以上のようにして決定された輪郭点が図14に示すように帯状領域の中心から離れていって帯幅の端まで達したときには、その段階で輪郭線が帯状領域から外れると判断して切り抜き画像の輪郭線を手動で設定するモードに自動的に切り換えるようにしても良い。この場合は、オペレータは帯状領域から外れた部分について画面上をポインティングデバイスにより輪郭線をトレースしていけばよく、再び輪郭線が帯状領域内に入ったときに自動モードに切り換えるようにすればよい。
【００７５】
以上示した各輪郭線抽出処理方法を含んだ実施例 (図９のＳ34，Ｓ35のサブルーチン) を図15に示す。
図において、まず、帯幅方向の輪郭点決定の重み分布を設定する（Ｓ61）。
次いで、帯状領域の帯幅方向の画像データに基づいて、輪郭点を算出し、複数の輪郭点候補を算出した場合は、前記重み分布に従って輪郭点を決定する（Ｓ62）。
【００７６】
そして、前記処理で輪郭点が算出されたか否か、即ち、前述のように差分の最大値のレベルが小さすぎて算出不能であるか否かを判定する（Ｓ63）。
ここで、算出不能であったと判別されたときには、前述した外挿処理を行って輪郭点の算出が不能であった区間の輪郭線を設定する（Ｓ64）。尚、算出不能であるときに手動モードに切り換えて、算出不能部分の輪郭点をポインティングデバイス操作によって手動で指定させるようにしても良い。
【００７７】
一方、輪郭点を算出できたときは、算出された輪郭点が前回の走査で決定された輪郭点の近傍に位置するか否かを判定し（Ｓ65）、離れている場合には前記外挿処理（或いは手動による輪郭点指定）を行い（Ｓ64）、近傍に位置する場合は、算出された輪郭点が帯状領域の幅方向の端に位置して、領域外に外れると予測されるか否かを判別する（Ｓ66）。
【００７８】
そして、領域から外れると判別されたときは、輪郭線を手動で設定するモードに切り換える（Ｓ67）。
次いで、前記手動モードによる輪郭線の設定が終了したか否かを判別し（Ｓ68）、終了した時点、つまり帯状領域内に戻った時点で自動抽出（自動探索）モードに切り換える（Ｓ69）。
【００７９】
尚、自動抽出された輪郭線が帯状領域から外れるときには、帯状領域の設定が不適切で、所望の輪郭線が帯状領域から外れている可能性が高いので、輪郭線が外れる方向に帯状領域の幅を自動的に所定範囲内で増大させるか、或いは自動的に湾曲させて輪郭点を算出させるようにしても良し、また、終点位置の指定のやり直しをオペレータに要求するようにしても良い。
【００８０】
一方、輪郭点が帯状領域から外れないと判定されたときは（Ｓ66）、算出された輪郭点を真の輪郭点として確定する（Ｓ70）。
そして、輪郭点の算出が終了したか否かを判別し（Ｓ71）、終了するまで上記の処理を繰り返す。
【００８１】
【発明の効果】
以上説明したように、請求項１の発明にかかる画像切り抜き装置によると、画像の切り抜き処理を行う領域を指定すると共に、前記処理領域での切り抜き条件を設定するための領域を別途指定し、前記切り抜き条件設定領域の画像データに基づいて条件設定して、切り抜き処理を行わせる構成としたので、画像個々の特性に適合する切り抜き条件を簡便に設定させることができ、以て、種々の原画像において安定的に画像の切り抜きを行うことができるという効果がある。
【００８２】
更に、切り抜き条件を複数の指定領域に基づいて設定させる構成とし、切り抜き条件の設定のために指定される領域に指定順に従って重み付けをすることで、例えば背景部分が変化するような主要画像において、複数の領域指定に基づく切り抜き条件の設定を適正に行うことができるという効果がある。
【００８３】
請求項２の発明にかかる画像切り抜き装置によると、種々の原画像において安定的に画像の切り抜きを行うことができると共に、切り抜き処理領域毎に異なる条件設定領域を対応させることで、各処理領域毎に切り抜き条件の適正が大きく変化する場合であっても、所望の画像切り抜きを行わせることができるという効果がある。
【００８４】
請求項３の発明にかかる画像切り抜き装置によると、種々の原画像において安定的に画像の切り抜きを行うことができると共に、切り抜き処理領域と条件設定領域とが重なるときに、該重複領域に重み付けして切り抜き条件を設定することで、切り抜き処理領域での画像特性により適合するように、切り抜き条件を設定させることができるという効果がある。
請求項４の発明にかかる画像切り抜き装置によると、切り抜き輪郭の輪郭線上で始点、終点を座標入力することで、簡便に輪郭線抽出を行わせるのに適正な帯状領域を切り抜き輪郭に沿って設定させることができ、然も、輪郭線抽出結果が、前記設定された領域と共に表示されて、オペレータの操作を容易にすることができるという効果がある。
【００８５】
請求項５の発明にかかる画像切り抜き装置によると、帯状領域の終点位置の更新設定毎に輪郭線抽出をやり直すので、抽出される輪郭を終点位置の指定によって意図的に誘導することができるという効果がある。
請求項６の発明にかかる画像切り抜き装置によると、帯状領域の帯幅方向で輪郭点を算出させるから、帯幅方向の中央に位置する可能性が高い所望の輪郭点を精度良く算出させることができると共に、かかる輪郭点の算出を帯状領域の始点から終点に向けて行わせることで、オペレータによる輪郭点の誘導が容易となるという効果がある。
【００８６】
請求項７の発明にかかる画像切り抜き装置によると、カラー原画像の輪郭を、色プレーン毎或いは色プレーン間の画像データ演算によって精度良く求めることができるという効果がある。
【図面の簡単な説明】
【図１】請求項１の発明にかかる装置の基本構成を示すブロック図。
【図２】請求項７の発明にかかる装置の基本構成を示すブロック図。
【図３】実施例のシステム構成を示すブロック図。
【図４】条件設定領域と切り抜き処理領域との指定の様子を示す図。
【図５】条件設定領域の指定による画像切り抜き処理の実施例を示すフローチャート。
【図６】条件設定領域の指定による画像切り抜き処理の別の実施例を示すフローチャート。
【図７】第２実施例の切り抜き処理の基本手順を示すフローチャート。
【図８】帯状領域の設定の様子と輪郭線との相関を示す図。
【図９】輪郭抽出の処理内容を示すフローチャート。
【図１０】帯状領域での輪郭点の抽出の様子を示す図。
【図１１】帯状領域の幅設定の様子を示すフローチャート。
【図１２】帯状領域を湾曲させる実施例を示すフローチャート。
【図１３】輪郭線の外挿の様子を説明するための図。
【図１４】手動モードへの切り換えの様子を説明するための図。
【図１５】輪郭線探索処理の詳細を示すフローチャート。
【図１６】分散度に基づく輪郭抽出の様子を示す図。
１ 画像入力手段
２ 演算手段
３ 内部記憶手段
４ 画像表示メモリ
５ ＣＲＴ
６ 入力手段
７ 外部記憶手段
８ マスク作成手段[0001]
[Industrial application fields]
The present invention relates to an image cropping device, and more particularly to an device for cropping a specific image on image data from image data obtained by photoelectrically converting an original image.
[0002]
[Prior art]
For example, in the plate making process for printing a product catalog, a product photograph used as a printing original image usually has an unnecessary background photographed around the product. Therefore, the background portion of the printing original image is erased (masked). A duplicate image needs to be obtained.
As a method of obtaining a duplicate image in which such a background portion has been erased, conventionally, a peel-off film is overlaid on a color separation plate, and an expert uses a light-shielding film ( There is a method of making a mask plate by cutting out a red film.
[0003]
However, in such a manual film cutting operation, it takes a long time to create a mask plate, and there is a problem that the finish of the mask plate varies depending on the operator.
In addition, there is a method of obtaining a desired mask image by extracting or erasing only a preset background color when an original image is read by a plate-making color scanner. Therefore, there is a possibility that accurate mask creation cannot be performed depending on the background setting.
[0004]
Further, the above-described manual clipping operation is electrically performed on the image data. This is because an original image of a product photograph or the like is read by photoelectric scanning, and the electrically read original image is displayed on a display, and an outline of the cutout is finely drawn by an operator using a pointing device such as a mouse or a trackball. A desired image is cut out on the image data according to the designated contour, and a duplicate image of only a necessary part is obtained.
[0005]
However, even in the above method, the cutter is merely changed to a pointing device, and basically a person cuts out. Therefore, simplification and automation of the cutting operation are desired.
In view of this, there has been proposed an apparatus that automatically obtains a contour line of a cutout region using density difference and color difference between an image region to be cut out and a background region, and cuts out a desired image based on the contour line ( (See Japanese Patent Publication No. 63-5745).
[0006]
[Problems to be solved by the invention]
However, when obtaining the outline of the cutout area based on the density difference or the color difference as described above, the original image to be cut out is various, and the difference between the area to be cut out and the background area is close Therefore, it has been difficult to stably cut out a desired image with high accuracy.
[0007]
In other words, the conventional configuration is such that the contour line of the clipped image is obtained from the image data in accordance with a certain contour extraction condition (clipping condition). For example, it is limited to cope with a change in the background setting of the original image. However, even if high-precision clipping was possible, stable automatic clipping was not possible with various original images.
The present invention has been made in view of the above problems, and the contour extraction conditions are determined according to the characteristics of each image. Appropriate It is an object of the present invention to provide an image cropping apparatus that can be easily set to 1 and to enable automatic image cropping processing to be stably performed on a wide range of original images.
[0008]
Also, in the image cropping process, only rough judgment / operation is left to the manual operation, and after that, the contour line of the cropped image can be automatically extracted with high accuracy so that the automatic cropping process of the stable image can be performed. For the purpose.
[0009]
[Means for Solving the Problems]
Therefore, claim 1 2 or 3 The image clipping device according to the invention is an image clipping device for clipping a specific image from image data obtained by photoelectrically scanning an original image, and is configured as shown in FIG.
In FIG. 1, a cutout area designating unit designates an area to cut out an image.
[0010]
On the other hand, the condition setting area designation means Specify the area for setting the cropping conditions. The clipping condition setting means sets the clipping condition based on the image data of the area specified by the condition setting area specifying means. Then, the image cutout means cuts out the image in the area specified by the cutout area specifying means in accordance with the cutout condition set by the cutout condition setting means.
[0011]
In the image clipping device according to the first aspect of the present invention, the clipping condition setting means weights each area according to a specified order based on the image data of a plurality of areas specified by the condition setting area specifying means. It was set as the structure which sets.
[0012]
In the image cutting device according to the second aspect of the invention, the image cutting means cuts out each of the plurality of areas specified by the cutting area specifying means based on each of the plurality of areas specified by the condition setting area specifying means. A plurality of clipping conditions set by the condition setting means are selectively applied.
[0013]
In the image cropping apparatus according to the invention of claim 3, when the clipping condition setting means has an area where the area specified by the clipping area specifying means and the area specified by the condition setting area specifying means overlap each other, The clipping condition is set by weighting the image data of the area.
On the other hand, the image clipping device according to the invention of claim 4 is configured as shown in FIG.
[0014]
In FIG. 2, the display means displays the original image based on the image data obtained by photoelectrically scanning the original image, and the contour coordinate designating means indicates the cutout contour on the image displayed on the display means. Start point and end point on contour Is specified.
Then, the band-shaped area setting means is designated by the contour coordinate designation means. Start and end points A band-like region along the cutout outline to be included is set, and the set band-like region is displayed on the display unit so as to overlap the image to be cut out.
[0015]
Outline extraction means Extracts the contour line of the image based on the image data in the belt-like region set by the belt-like region setting means, and the contour line display means Outline extraction means The contour line extracted in step (1) is displayed on the display unit in the band-like region so as to overlap the image to be cut out. And the image cropping means Outline extraction means The image is cut out based on the contour line extracted in step (1).
[0016]
In the image cropping device according to the fifth aspect of the invention, the contour line extracting unit changes the contour of the image in the band-shaped region every time the coordinate position is newly specified by the contour coordinate specifying unit and the end point position of the band-shaped region is changed. It was set as the structure newly extracted again.
In the image cropping device according to the invention of claim 6, the contour line extracting means extracts a contour point on the contour line in the image data sequence in the band width direction of the band-like region, and for each image data sequence. The contour points are extracted sequentially from the start point position to the end point position of the band-like region according to the coordinate position designation order by the contour coordinate designation means.
[0017]
In the image cropping device according to the seventh aspect of the invention, the image data is color image data, and the contour line extraction unit extracts a contour line by calculation between color planes or for each color plane.
[0018]
[Action]
According to the image cropping apparatus of the first aspect of the present invention, the region for cropping the image and the region for setting the cropping conditions are individually set. Here, by specifying the region to be clipped, it is possible to easily extract the image to be clipped by eliminating images other than the clip target as much as possible, and for example, specify the background portion of the clip target image as the region for condition setting Therefore, it is possible to perform cropping under conditions suitable for the characteristics of each image.
[0019]
Furthermore, the degree of contribution to the setting of the clipping condition can be ranked for each region by setting the clipping condition by weighting according to the region designation order based on the image data of a plurality of regions set for condition setting. Thus, it is possible to set appropriate clipping conditions comprehensively determined from the plurality of areas.
[0020]
In the apparatus according to the second aspect of the present invention, a plurality of cropping conditions set based on a plurality of designated regions can be selectively applied to each of a plurality of regions designated as regions for image clipping. In each of the plurality of cutout areas, the image can be cut out under appropriate conditions.
[0021]
In the apparatus according to the third aspect of the present invention, when there is an area where the area designated as the area to be clipped and the area designated for setting the clipping condition are mutually overlapped, the area is designated for setting the clipping condition. In particular, the image data of the portion that overlaps the cutout area is regarded as image data suitable for setting the cutout condition, and the cutout condition is set by weighting the image data in the overlapped area. Make it.
[0022]
On the other hand, the image cropping apparatus according to the invention of claim 4 displays the target image of the cropping process, and designates the start point and the end point on the contour line of the contour of the image to be clipped while viewing the display. When the start point and the end point are designated, a band-like area is set along the cut-out area including the start point and end point, and the band-like area is displayed so that the setting state of the band-like area can be confirmed. Here, the contour line of the image is extracted based on the image data in the belt-like region, and the extracted contour line is displayed together with the display of the belt-like region so that the automatically extracted contour line can be visually confirmed. . Finally, the image is cut out according to the extracted contour line.
[0023]
In other words, when the outline of the image to be clipped is roughly specified, the outline extracted automatically within the specified area is displayed on the image together with the specification, and the coordinate designation of the outline and the automatic extraction result are visually confirmed. However, it is possible to perform coordinate designation work. In the apparatus according to the fifth aspect of the invention, every time the end position of the belt-like region is changed by newly designating the coordinate position, contour extraction is performed again, and the contour line automatically extracted by the designation of the coordinate position is the desired one. To be able to guide.
[0024]
In the apparatus according to the sixth aspect of the present invention, it is predicted that the coordinate position is designated so that the contour line is located in the vicinity of the approximate center in the band width direction. Contour extraction is facilitated as a configuration for extracting points. Further, the extraction of the contour points for each image data sequence is sequentially performed from the start point position to the end point position of the band-like region according to the coordinate position designation order, and the contour lines are sequentially extracted and displayed along the coordinate position designation order. So that
[0025]
In the apparatus according to the seventh aspect of the invention, when the image data is color image data, the contour line is extracted by the difference in characteristics between the background color and the color of the image to be clipped by calculation between the color planes or for each color plane. It was made to be.
[0026]
【Example】
Examples of the present invention will be described below.
FIG. 3 is a block diagram showing the hardware configuration of this embodiment. The image input means 1 obtains color separation image data by photoelectrically scanning an original image such as a product color photograph. The color separation image data obtained by the above is temporarily stored in the internal storage means 3 such as a RAM via the calculation means 2.
[0027]
The color separation image data is also stored in the image display memory 4, and the original image is displayed on the CRT 5 (display means) using the image display memory 4.
The calculation means 2 serving as a clipping condition setting means and an image clipping means extracts a contour of an image (main image) to be cut out based on the color separation image data, and duplicates a main image from which unnecessary portions such as a background are deleted. Data is set, and a mask for printing is created by the mask creating means 8 based on the duplicate image data.
[0028]
Input means 6 that is a man-machine interface is connected to the calculation means 2, and the operator can give various image cutout information and instructions via the input means 6.
In the present embodiment, the input means 6 is a clipping area designating means for designating an area for image clipping and an image clipping condition (contour extraction processing condition) in the designated area. It also serves as condition setting area designating means for designating an area.
[0029]
That is, the input means 6 includes a pointing device such as a mouse or a trackball, and the operator wants to execute a clipping process while viewing the original image displayed on the CRT 5 (an area including a cutting outline). : Mask area) is specified by inputting the coordinate position by the pointing device, and an area (condition setting area) suitable for condition setting in the automatic extraction of the contour line in the area is also set to the coordinate position by the pointing device. Designate by input (see FIG. 4).
[0030]
In this way, by specifying roughly the region including the contour portion of the image to be clipped by the designation by the input means 6 as the clip region designation means, the image data to be subjected to automatic contour extraction is reduced as much as possible, and the contour extraction is performed. Can be performed accurately in a short time. Furthermore, in automatic contour extraction based on image data calculation processing, how much density difference and color difference should be considered as the contour of the image to be cut out depends on the background setting of each image and is constant. It is difficult to extract a stable contour line under the above-mentioned clipping conditions, and it is difficult for an operator to accurately indicate the contour extraction conditions (cutout conditions) such as density differences and color differences as numerical values. Therefore, in this embodiment, the operator designates the background portion of the image to be cut out as the condition setting area, and sets the cut-out condition according to the designation, so that the condition according to the setting of the background portion of each image is set. Thus, the image can be stably cut out.
[0031]
With this configuration, the computing means 2 extracts the contour of the main image from the color separation image data, and the image cropping result by the computing means 2 is stored / saved in the external storage means 7 comprising a hard disk or optical disk, while creating a mask. Provided to means 8.
The mask creating means 8 fills unnecessary portions other than the cut-out image portion in accordance with the cut-out image data, enlarges / reduces the mask data, and creates mask data used for the plate collecting operation (such mask data). May be used to make a mask film).
[0032]
FIG. 5 is a flowchart showing the procedure of the image clipping process performed by the above configuration.
In the flowchart of FIG. 5, first, the operator designates an area for sampling image data as a basis for condition setting in automatic contour extraction, that is, a condition setting area via the input means 6 (S1: condition setting area specifying means). .
[0033]
In designating the condition setting area, the operator designates an area including only the background portion of the main image with a pointing device while viewing the original image displayed on the CRT 5 (see FIG. 4). Specifically, as shown in FIG. 4, a rectangular area can be specified by specifying the coordinates of the apex portion with a pointing device, and such a specification result is displayed on a line drawing superimposed on the original image displayed on the CRT 5 Good to do.
[0034]
As shown in FIG. 4, when a double area is designated within the area designated as the condition setting area, an annular area surrounded by two area sections is set as the condition setting area. Thus, when an image not suitable for the condition setting is included in the background portion, it is preferable that the portion can be easily removed.
The condition setting area may be specified by one, or a plurality of areas may be specified individually or partially overlapping. When the background portion is not constant but shows a change, by specifying a plurality of representative regions of the changing background portion as the condition setting region, it is possible to set an optimum clipping condition in the condition setting process described later.
[0035]
In addition, when multiple areas are specified as condition setting areas, there is a possibility that areas that are to be greatly contributed to condition setting and areas that are desired to be stopped for reference may be divided. For example, the condition setting areas are weighted according to the specified order. Then, image data included in an area with a fast designation order should have a great influence on the condition setting, and the operator should be able to select whether or not to perform weighting according to such designation order.
[0036]
The specification of the condition setting area and the clipping process area (mask area) to be described later is not limited to specifying the coordinate position of the apex portion of the square area with the pointing device as described above, and a digitizer, a tablet, etc. It may be used so that it can be specified in an arbitrary shape.
When the specification of the condition setting area is completed, the calculation means 2 samples the image data in the area specified as the condition setting area, and in other words, the sampled image data, in other words, the image data of the background portion of the main image. Based on the above, the clipping condition (mask creation condition) of the main image is set (S2: clipping condition setting means).
[0037]
For the setting of the clipping condition, for example, an average value (r + g + b) / 3 of color separation image data (for example, R, G, B three primary color signals r, g, b) in the condition setting region is obtained, and the average is obtained. Using a value as a threshold value, a pixel in a clipping processing area, which will be described later, is discriminated from a background pixel and a main image pixel.
Thereby, discrimination (contour extraction) between the pixels of the background portion and the pixels of the main image can be performed with high accuracy according to the characteristics of the background portion of each image. That is, in contour extraction under a certain condition, it is difficult to stably cut out because it cannot cope with changes in the background part, but if the background part image is specified by specifying the condition setting area as described above, It becomes easy to discriminate the specified background portion in the region where the clipping process is performed, so that it is possible to perform the clipping with high accuracy for a wide range of images.
[0038]
When the clipping condition (image signal discrimination threshold or the like) is determined based on the designated condition setting area, such a condition is displayed on the CRT 5 and whether or not the operator's desired clipping condition is set. May be determined.
Further, it is obvious that the setting of the clipping condition is not limited to the average value of the above image signal. For example, the color image data included in the condition setting area is averaged for each color plane, and based on this A threshold value may be set for each color, or a threshold value may be set based on image data having the highest frequency in the condition setting area instead of the average value. Furthermore, the main color of the background portion (condition setting region) can be specified from the histogram information for each color, and only the main color can be determined for contour extraction.
[0039]
Further, the dispersion degree σ of the color separation data in the condition setting area is obtained according to Equation 1, and the dispersion degree σ is set as a threshold value, while the cut-out processing area is scanned in a small area and included in the small area at the scan position. Alternatively, the degree of dispersion σ of the image data to be obtained may be obtained, and the center of the region having the degree of dispersion σ greater than or equal to the threshold may be extracted as a contour (see FIG. 16).
[0040]
[Expression 1]

[0041]
When the designation of the condition setting area and the setting of the clipping condition based on the designated area are completed (S3), the input means 6 designates the area (mask area) where the image is actually cut out next. (S4: Clipping area designation means).
In specifying the cut-out processing area, an area that includes the boundary between the main image and the background part is specified by the coordinate specification by the pointing device while the operator looks at the original image on the CRT 5 as in the condition setting area specification. The set clipping processing area is displayed so as to overlap the original image on the CRT 5.
[0042]
In specifying the processing area, it is preferable to specify the boundary portion divided into a plurality of areas as shown in FIG. 4 because reducing the area as much as possible reduces the burden of the automatic contour extraction processing. In order to include the contour portion in the processing region without omission, it is preferable to allow overlap between designated processing regions, and finally, a region including a plurality of designated processing regions is set as the processing region.
[0043]
The specified condition setting area and the processing area specification may partially overlap, or the processing area may be specified in a form including the condition setting area.
Further, the designation of the area to be cut out is not limited to the square area as shown in FIG. 4, and the area designation by an arbitrary shape may be performed.
[0044]
Further, when the main image to be cut out is small, an area to be cut out may be designated so as to include the main image.
When the area specification for the clipping process is completed, the outline of the main image is automatically extracted in the clipping process area in accordance with the set clipping condition (S5: image clipping means), and the automatic extraction process is normally performed. (S6), the extraction result is finalized, and thus a mask is created (S5).
[0045]
Specifically, when the clipping condition is set as the image data threshold value from the image data in the condition setting area, the image data in the designated processing area is compared with the threshold value, and the pixel corresponding to the main image By dividing into pixels corresponding to the background portion, the outline of the main image is extracted.
That is, in the above-described embodiment, a single clipping condition is set in a plurality of processing areas based on image data in a plurality of designated condition setting areas or in one condition setting area. A clipping process is executed.
[0046]
Therefore, when the clipping condition is set based on a plurality of condition setting areas, the background portion that changes is comprehensively determined. Appropriate In addition, it is possible to set a clear cutting condition, and further, by using one cutting condition in common for a plurality of processing areas, a uniform cutting process can be performed in each processing area.
Here, the extracted contour line may be displayed on the CRT 5 so as to overlap the original image so that the operator can determine whether or not the automatic extraction result is correct. If there is a defect in the automatically extracted contour, it is possible to manually correct the contour with a pointing device, to manually correct the clipping conditions, or to respecify the condition setting area. It is good to be able to.
[0047]
In the procedure shown in the flowchart of FIG. 5 above, the condition setting area is specified first, and then specified in accordance with the clipping condition determined based on the image data included in the specified condition setting area (single or plural). However, as shown in the flowchart of FIG. 6, the clipping process areas are specified first, and the conditions of the clipping process in these areas are selected. The condition setting area for setting can be designated later.
[0048]
In the flowchart of FIG. 6, first, the operator designates the clipping process region (plurality or singular number) including the contour portion of the main image (see FIG. 4) using the pointing device in the same manner as in the flowchart of FIG. 5. (S11: Clipping area designating means).
When the specification of the processing area is completed so as to cover all the contours of the main image (S12), a plurality or a single designation of condition setting areas for setting the clipping processing conditions in the designated processing area is subsequently performed ( As shown in the flowchart of FIG. 5, the operator uses a pointing device (S13: condition setting area designating unit).
[0049]
Here again, setting of a condition setting area that is included in the processing area (designation of a background part in the processing area) or setting of a condition setting area that partially overlaps the processing area is allowed. good.
When the designation of the condition setting area (multiple area or single area) is completed, the main image is cut out based on the image data of the designated condition setting area, that is, the image data corresponding to the background portion of the main image (contour extraction) ) Is set (S14: Clipping condition setting means).
[0050]
When a plurality of condition setting areas are set, the cutting conditions may be set by weighting the areas in accordance with the designation order as described above.
Further, in the procedure shown in the flowchart of FIG. 6, since the processing area is already specified when the condition setting area is specified, the correlation between the condition setting area and the processing area is specified when the clipping condition is set. It will be. Therefore, when the processing region and the condition setting region overlap, the image data of the overlapping region is a background portion close to the contour portion of the main image, so the image of the overlapping region is compared with the image data of the overlapping region. It is preferable to weight the data so that the image data in the overlapping area has a great influence on the setting of the clipping condition.
[0051]
When the setting of the cutout conditions is completed, the outline of the main image is extracted in each designated processing area according to the actual conditions, and mask creation is performed (S15: image cutout means).
In addition, after specifying a plurality of cropping processing areas, or each time specifying a cutting processing area, specify and specify an area for setting a cutting condition for each processing area, or specify a plurality of condition setting areas Or each time a condition setting area is designated, an area to be clipped is specified and indicated by a clipping condition based on each condition setting area. For example, processing area A is based on image data in condition setting area a. The cutting condition may be set, and the processing area B may be set based on the image data of the condition setting area b so that a different condition setting area is associated with each processing area.
[0052]
According to such a configuration, even when the background portion of the main image changes greatly depending on the part, it is possible to accurately extract the corresponding contour portion under a plurality of clipping conditions according to the background change. Become.
Further, for example, for the processing area A, a clipping condition is set based on the image data of the condition setting areas a and c, and for the processing area B, a clipping condition is set based on the image data of the condition setting areas b and c. Although the combination of condition setting areas corresponding to each processing area is different, it is possible to have a configuration in which partially overlapping condition setting areas exist.
[0053]
In addition, when processing areas and condition setting areas are displayed on the CRT 5, each area is given an identification code to facilitate confirmation of the correspondence between the processing area and the condition setting area. It is preferable to make it easy.
By the way, as described above, when the operator roughly specifies the region including the contour portion of the cutout, the contour of the main image is aligned in order to reduce the burden of the automatic contour extraction processing and improve the accuracy. It is desirable to be able to specify a narrow area as easily as possible. Further, it is desired that the contour extraction is performed reflecting the operator's intention to specify the region, and the result of the contour extraction can be easily confirmed following the region specification.
[0054]
Therefore, a more preferred embodiment of the region designation and contour extraction of the contour portion will be described below. The second embodiment described below will be described with reference to the hardware configuration shown in FIG. 3 in the first embodiment described above.
The flowchart of FIG. 7 shows the basic procedure of mask creation in the second embodiment. First, the operator views the original image on the CRT 5 (display means) and inputs the main image via the input means 6 (pointing device). One point on the contour line is input as the coordinate position of the starting point for performing contour extraction (S21: contour coordinate designating means). Subsequently, similarly, the input unit 6 inputs coordinates of one point on the contour line of the main image different from the start point as an end point position for performing contour extraction (S22: contour coordinate designating unit).
[0055]
By inputting the coordinates of the start point / end point position, as shown in FIG. 8, a belt-like region along a cutout outline that linearly connects a predetermined radius range centered on the start point and a predetermined radius range centered on the end point. Is set as a contour extraction region (cutout region), and such a band-like region is displayed as a line drawing on the CRT 5 so as to overlap the original image. Then, contour extraction is performed within the extraction region by the computing means 2, and the extracted contour is also displayed superimposed on the image on the CRT 5 (S23: strip region setting means, contour line extraction means, contour line display means) ).
[0056]
That is, in the first embodiment, an area is specified so as to surround the outline, but in the second embodiment, an outline extraction area including the outline is set by specifying a main point on the outline. The operator can trace the outline of the main image and specify the main points on the outline with a pointing device such as a mouse. However, in the band-like region in the present embodiment, it is highly possible that the contour line of the main image is located near the center in the width direction of the band-like region, so that the contour extraction from the band-like region can be easily performed.
[0057]
Until a series of contour point extraction is completed by setting the band-like region (contour extraction region) accompanying the input of the coordinate position of the principal point on the contour of the principal image as described above (S24), the coordinate position is determined by the input means 2. The input points are sequentially updated and set as end point positions, and contour extraction is performed (contour line extraction means).
That is, when a point on the contour line of the image to be clipped is first specified, it is first set as the start point, and then when a different point on the contour line is specified, a band-like region connecting the start point and the end point is set. Contour extraction is performed within the region. Subsequently, when a new point on the contour line is designated, a band-like area is extended from the point on the outline previously designated as the end point position to the new designated point (end point), and within the extended band-like area. Then, the contour extraction is performed, and the coordinate input of the points on the contour line is repeated until the setting of the band-like region covering the series of contour lines is completed.
[0058]
Here, when there is an update setting for the end point position, it is preferable to cancel the contour obtained up to the previous time and newly perform contour extraction within the entire belt-like region. This is because a plurality of contour point candidates are detected in contour extraction, and when one contour point is finally selected from these, the update setting of the end point position suggests the direction of the contour line, This is because the contour candidate point to be selected is changed. In the direction of contour extraction by specifying the end point position, the extracted contour points are displayed on the CRT 5, so that the operator guides the contour line extracted based on the display of the contour points in a desired direction. Thus, it is possible to intentionally specify the end point position.
[0059]
When a series of contour lines are extracted, a new starting point for contour extraction is designated (S25), the band-like region is set and the contour extraction is performed in the same manner, and the above processing is performed until all contour extraction is completed. Is repeated (S26).
When all contour extraction is completed, the extracted contour points are vectorized (each contour point is connected) to create a mask image (S27: image cutout means).
[0060]
Next, the outline extraction in S23 of the flowchart of FIG. 7 will be described in detail with reference to the flowchart of FIG.
First, as described above, based on the main points (start point and end point) coordinate-input on the contour of the image desired to be clipped, a strip region for detecting the contour point is set (S31: strip region setting means). .
[0061]
Next, among the color image data (R, G, B or Y, M, C, K) obtained by photoelectric conversion of the original image, the image data in the band-like region is converted into H (hue) by a known method. , S (saturation), and L (lightness) data (S32).
Then, normalization constants of a plurality of image data used for contour point extraction are set (S33). Specifically, for example, the maximum value X of each image data is set so that the width of the maximum value and the minimum value of each image data is the same. _max And minimum value X _min X = X / (X _max -X _min ) × 100 (where X = R, G, B, H, S, L).
[0062]
Subsequently, the image data is scanned from the start point side in the width direction intersecting with the center line connecting the start point and the end point of the band-like region (see FIG. 10), and the contour line of the main image on the scan line (on the image data string) Contour point candidates that intersect with (S34: contour line extracting means). Specifically, for each image data (each color plane), a difference in data between adjacent pixels on the scanning line (image data string) is obtained, and a contour candidate point is determined as a pixel position where the difference is equal to or greater than a predetermined value. Ask.
[0063]
Note that the extraction of contour points is not limited to the configuration using the image data R, G, B, H, S, L, and instead of calculating data changes between pixels for each color plane, It is also possible to obtain a contour point by obtaining a data deviation between color planes.
Further, in the same way as in the first embodiment, in addition to the designation of the band-like area (cutout process area), the area (condition setting area) including the background portion of the main image is designated, and the image included in the designated area is designated. A threshold value (cutout condition) used for calculating the contour point on the scanning line may be set based on the data.
[0064]
The contour points are calculated at regular intervals (one pixel unit or a plurality of pixel lengths) from the start point to the end point every time the end point position is set, and the contour candidate points obtained on each scanning line are consecutive. The connectivity (connectivity) is evaluated, and contour candidate points on each scanning line are selected from the start point to the end point (S35). Then, the selected contour candidate points are vectorized (each contour point is connected) to obtain a contour line (S36), and the contour line is superimposed on the original image and displayed on the CRT 5 (display means) (S37). : Outline display means).
[0065]
The operator can visually confirm the contour line extracted based on the designated start point and end point position by the display on the CRT 5, thereby determining the suitability of the automatically extracted contour line, and thereafter Since the contour extraction result can be predicted, the displayed extraction result can be used as an index for updating the end point position, and the extracted contour line can be set in a desired direction by specifying the end point position. It is possible to guide.
[0066]
In other words, it can be determined from the extracted contour extraction result that it is possible to trace a point that is different from the original contour line to be cut out or that there is a high possibility of tracing an erroneous contour part. In the new designation of the end point position, the part that is erroneously traced or the part that is likely to be erroneously traced is excluded from the band-like area, or the interval of the designated position of the end point position is changed. Thus, it is possible to devise so that the intended contour line is traced. Since the detection of the contour point is performed again from the start point every time the end point position is designated, the continuity evaluation changes depending on the setting of the new end point position, and the intended contour line can be traced.
[0067]
Next, a more preferable aspect based on the cutout process control will be described.
First, as shown in FIG. 8, the shape of the band-shaped region set based on the designation of the start point and end point position is a circular outline when the center of the circle is moved between the set start point and end point. The trajectory shape is basically set to a shape in which two circles are connected in a straight line. With such a region, the contour extraction region is set to have a constant width (circular diameter) regardless of the direction of the belt-like region, so that the contour line extraction accuracy can be kept constant. Here, the width of the band-like region is preferably set variably, and the outline extraction accuracy can be increased by automatically setting as follows.
[0068]
For example, as the band width setting method, the band width is set to a predetermined ratio (pixel number ratio), for example, 1/20, with respect to the size of the displayed clipped target image. Since the spatial change of the pixel value for displaying the image structure of the image differs depending on the reading resolution of the image, the band width is set so as to be a fixed ratio with the image clipping main body. For example, the arrangement of adjacent pixel values constituting the same contour line is greatly different between an image read at the same magnification and an image read at 400%. In this embodiment, an appropriate band width is selected from the image area to be cut out and the image size specified by the operator (1:20). Note that the pixel ratio is preferably 1:15 to 1:25. Below this, the estimated contour line is likely to be affected by scratches and other noise, and above this, it becomes difficult to extract an appropriate contour point. In particular, 1:18 to 1:22 is preferable. In this case, since the size (number of display pixels) of the displayed clipped image cannot be obtained without knowing the outline of the image, for example, a rectangular region that is substantially circumscribed by the clipped image is set, and the size of the region ( The band width may be set based on the number of pixels.
[0069]
FIG. 11 shows such an embodiment (subroutine S31 in FIG. 9).
In the figure, first, the display size S of the cut-out image is read (S41), then the band width W is set as 1/20 · S (pixel number ratio) (S42), and the start point is set with the set band width W. A band-like area connecting the end point is set, and the band-like area is displayed on the CRT 5 (S43).
[0070]
In addition, it is preferable that the band width W can be set manually in consideration of the fact that there is a portion where the change rate of the contour line locally increases even in the same size image.
By the way, if the belt-like region is set to only a linear shape, it is necessary to set the start point and the end point many times by connecting the start point and the end point to the contour line having a large curvature. Therefore, by curving and setting the belt-like region according to the operator's instruction, a long contour line can be included in one belt-like region, and the number of times of setting can be reduced. As a specific curving method, for example, a Bezier function can be used.
[0071]
FIG. 12 shows such an embodiment (subroutine S31 of FIG. 9).
In the figure, first, the band width W is set by the above-described method or the like (S51), and then the operator's instruction as to whether or not the band-like region is to be curved is read (S52). When the area is displayed (S54) and curved, the bending process using the Bezier function or the like is performed (S53), and the band-shaped area subjected to the bending process is displayed on the CRT 5 (S55). Then, the band-shaped region that has been curved by the operator Appropriate (S56), the bending process is continued until it becomes OK, and if it becomes OK, the curved band-like region is finally set (S57).
[0072]
On the other hand, when a plurality of contour candidate points are detected on the same scanning line, as a method of determining a truly desired contour point from these, a weight is applied in advance in the width direction (scanning line direction) of the band-like region. It is preferable to keep it.
That is, when the operator sets the band-like area, considering that the probability that the center line of the band-like area is unintentionally set close to the desired cutout outline is high, the image data near the center in the width direction of the band-like area is set. The weight is set larger, and the weight is set smaller at the end, and when there are a plurality of contour point candidates, the larger weight is selected. However, there is a high probability that the outline is near the center of the band width at the start point and end point, but the possibility of moving away from the center increases in the middle part, so weighting only the part near the start point and end point, or the degree of weighting Or you may make it change with the distance from an end point.
[0073]
Further, when the density or color change in the band width scanning direction is small and it is impossible to determine the contour point from the difference between the data, it is possible to follow the extending direction of the band-like region for such an indistinguishable portion. As shown in FIG. 13, the contour line is determined by extrapolation. Further, it is determined whether or not the determined contour point is in the vicinity of the contour point determined on the previous scanning line. If it is not in the vicinity, it is determined that the determination of the contour point is incorrect and the same as described above. If the contour point is determined by insertion, or if there are multiple contour point candidates on the same scanning line, the second candidate is compared with the previously determined contour point, and if it is in the vicinity, the first candidate is determined Alternatively, the second candidate may be selected again.
[0074]
Furthermore, when the contour point determined as described above is far from the center of the strip region and reaches the end of the strip width as shown in FIG. 14, it is determined that the contour line is out of the strip region at that stage. Thus, the mode may be automatically switched to a mode in which the contour line of the cutout image is manually set. In this case, the operator only needs to trace the contour line on the screen by using a pointing device for the part outside the belt-like region, and switch to the automatic mode when the contour line again enters the belt-like region. .
[0075]
FIG. 15 shows an embodiment (subroutines S34 and S35 in FIG. 9) including the contour extraction processing methods shown above.
In the figure, first, a weight distribution for determining contour points in the width direction is set (S61).
Next, contour points are calculated based on image data in the band width direction of the band-like region, and when a plurality of contour point candidates are calculated, the contour points are determined according to the weight distribution (S62).
[0076]
Then, it is determined whether or not the contour point has been calculated in the process, that is, whether or not the maximum difference level is too small to be calculated as described above (S63).
If it is determined that the calculation is impossible, the extrapolation process described above is performed to set the contour line of the section where the contour point cannot be calculated (S64). Note that when the calculation is impossible, the mode may be switched to the manual mode so that the contour points of the non-calculated part can be manually designated by operating the pointing device.
[0077]
On the other hand, when the contour point can be calculated, it is determined whether or not the calculated contour point is located in the vicinity of the contour point determined in the previous scan (S65). Processing (or manual contour point designation) is performed (S64), and if it is in the vicinity, whether or not the calculated contour point is located at the end in the width direction of the belt-like region and is predicted to be out of the region Is determined (S66).
[0078]
When it is determined that the region is out of the region, the mode is switched to a mode for manually setting the contour line (S67).
Next, it is determined whether or not the setting of the contour line in the manual mode has been completed (S68), and when it is completed, that is, when it returns to the belt-like region, the mode is switched to the automatic extraction (automatic search) mode (S69).
[0079]
When the automatically extracted contour line deviates from the band-like region, the setting of the band-like region is inappropriate, and there is a high possibility that the desired contour line has deviated from the band-like region. The width may be automatically increased within a predetermined range, or the contour point may be calculated by being automatically curved, or the operator may be requested to redesignate the end point position.
[0080]
On the other hand, when it is determined that the contour point does not deviate from the belt-like region (S66), the calculated contour point is determined as a true contour point (S70).
Then, it is determined whether or not the calculation of the contour point has been completed (S71), and the above processing is repeated until the calculation is completed.
[0081]
【The invention's effect】
As described above, according to the image cropping apparatus according to the invention of claim 1, the region for performing the cropping process of the image is designated, and the region for setting the cropping condition in the processing region is separately designated, Since it is configured to perform the cutting process by setting conditions based on the image data in the cutting condition setting area, it is possible to easily set cutting conditions suitable for the characteristics of each image, and thus various original images In this case, the image can be stably cut out.
[0082]
Furthermore, in the main image in which, for example, the background portion changes, the cropping conditions are set based on a plurality of designated areas, and the areas designated for setting the clipping conditions are weighted according to the designation order. There is an effect that it is possible to appropriately set the clipping conditions based on a plurality of area designations.
[0083]
According to the image cropping apparatus of the second aspect of the present invention, it is possible to stably crop an image in various original images, and to associate different condition setting regions for each cropping processing region, so that each processing region Even if the appropriateness of the cropping condition changes greatly, there is an effect that desired image cropping can be performed.
[0084]
According to the image cropping device of the third aspect of the invention, it is possible to stably crop an image in various original images, and when the cropping processing area and the condition setting area overlap, the overlapping area is weighted. By setting the clipping condition, it is possible to set the clipping condition so as to be more suitable for the image characteristics in the clipping processing area.
According to the image cropping apparatus of the invention of claim 4, by setting the coordinates of the start point and end point on the contour line of the cutout contour, an appropriate belt-like region can be set along the cutout contour for easy contour line extraction. However, the contour line extraction result is displayed together with the set area, so that the operator can easily operate.
[0085]
According to the image cropping apparatus of the fifth aspect of the present invention, the contour line extraction is performed again every time the end point position of the belt-like region is updated, so that the extracted contour can be guided intentionally by specifying the end point position. There is.
According to the image cropping device of the sixth aspect of the present invention, the contour point is calculated in the band width direction of the band-like region, so that it is possible to accurately calculate a desired contour point that is likely to be located in the center of the band width direction. In addition, the calculation of the contour point is performed from the start point to the end point of the belt-like region, so that the operator can easily guide the contour point.
[0086]
According to the image cropping device of the seventh aspect of the invention, there is an effect that the contour of the color original image can be obtained with high accuracy by image data calculation for each color plane or between color planes.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a basic configuration of an apparatus according to the invention of claim 1;
FIG. 2 is a block diagram showing a basic configuration of an apparatus according to the invention of claim 7;
FIG. 3 is a block diagram showing a system configuration of the embodiment.
FIG. 4 is a diagram showing how a condition setting area and a clipping process area are specified.
FIG. 5 is a flowchart illustrating an example of image clipping processing by specifying a condition setting area.
FIG. 6 is a flowchart illustrating another example of image cropping processing by specifying a condition setting region.
FIG. 7 is a flowchart showing a basic procedure of a clipping process according to the second embodiment.
FIG. 8 is a diagram showing a correlation between a setting state of a band-like region and an outline.
FIG. 9 is a flowchart showing processing details of contour extraction.
FIG. 10 is a diagram showing how contour points are extracted from a belt-like region.
FIG. 11 is a flowchart showing how to set the width of a belt-like region.
FIG. 12 is a flowchart showing an embodiment for bending a belt-like region.
FIG. 13 is a diagram for explaining a state of extrapolation of a contour line.
FIG. 14 is a diagram for explaining a state of switching to a manual mode.
FIG. 15 is a flowchart showing details of contour search processing;
FIG. 16 is a diagram showing a state of contour extraction based on the degree of dispersion.
1 Image input means
2 Calculation means
3 Internal storage means
4 Image display memory
5 CRT
6 Input means
7 External storage means
8 Mask making means

Claims (3)

An image clipping device for clipping a specific image from image data obtained by photoelectrically scanning an original image,
A cropping area designating means for designating an area for cropping an image;
A condition setting area specifying means for specifying an area for setting a cutting condition;
Clipping condition setting means for setting a clipping condition based on the image data of the area specified by the condition setting area specifying means;
Image cutting means for cutting an image in the area specified by the cutting area specifying means according to the cutting conditions set by the cutting condition setting means;
Comprising
The image cropping condition setting unit sets the cropping condition by weighting each region in accordance with a specified order based on image data of a plurality of regions specified by the condition setting region specifying unit. apparatus. An image clipping device for clipping a specific image from image data obtained by photoelectrically scanning an original image,
A cropping area designating means for designating an area for cropping an image;
A condition setting area specifying means for specifying an area for setting a cutting condition;
Clipping condition setting means for setting a clipping condition based on the image data of the area specified by the condition setting area specifying means;
Image cutting means for cutting an image in the area specified by the cutting area specifying means according to the cutting conditions set by the cutting condition setting means;
Comprising
A plurality of image clipping means set by the clipping condition setting means based on each of the plurality of areas designated by the condition setting area designation means for each of the plurality of areas designated by the clipping area designation means. An image cropping apparatus characterized by selectively applying the cropping condition of. An image clipping device for clipping a specific image from image data obtained by photoelectrically scanning an original image,
A cropping area designating means for designating an area for cropping an image;
A condition setting area specifying means for specifying an area for setting a cutting condition;
Clipping condition setting means for setting a clipping condition based on the image data of the area specified by the condition setting area specifying means;
Image cutting means for cutting an image in the area specified by the cutting area specifying means according to the cutting conditions set by the cutting condition setting means;
Comprising
When the clipping condition setting means has an area where the area specified by the clipping area specifying means and the area specified by the condition setting area specifying means overlap each other, the image data of the overlapping area is weighted. An image cropping apparatus characterized by setting cropping conditions.

Images