CN101436314B - Brake machine channel system, method and system for recognizing passing object in brake machine channel - Google Patents
Brake machine channel system, method and system for recognizing passing object in brake machine channel Download PDFInfo
- Publication number
- CN101436314B CN101436314B CN2007100313843A CN200710031384A CN101436314B CN 101436314 B CN101436314 B CN 101436314B CN 2007100313843 A CN2007100313843 A CN 2007100313843A CN 200710031384 A CN200710031384 A CN 200710031384A CN 101436314 B CN101436314 B CN 101436314B
- Authority
- CN
- China
- Prior art keywords
- output state
- current target
- gait
- gate passage
- collects
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Landscapes
- Image Analysis (AREA)
Abstract
The embodiment of the invention discloses a method for identifying a traffic target in a gate passage. The method comprises the following steps: an output state of a gait sensor in the gate passage is acquired; when a disconnected output state output by the gait sensor is acquireed, the gate passage is subjected to video acquisition; and according to the acquired information of the output state of the gait sensor and the acquireed video information, the class, the number, the movement direction, the movement speed and the ascription relation of the traffic target in the gate passage are identified. The embodiment also discloses a system for identifying the traffic target in the gate passage and a gate passage system. The method for identifying a traffic target in a gate passage has the advantage of identifying the class, the number, the movement direction, the movement speed and the ascription relation of the traffic target in the gate passage through a simple, accurate and reliable method.
Description
Technical field
The present invention relates to the ticket-checking system field, relate in particular to the current target identification method and the system of a kind of gate channel system, gate passage.
Background technology
At present, growing along with traffic system, the gate automatic ticket checking system is progressively adopted in each country, area, and the gate ticket-checking system of prior art adopts the mode of placement sensor in the gate passage quick ticket checking to be provided and to come in and go out current service for the passenger usually.The scheme of prior art only can simply be discerned object current in the passage, making simple current relation, can not handle the current relation of complexity (current etc. with goods such as adjacent current, the pedestrian of, many people).
Summary of the invention
Given this, the embodiment of the invention provides the current target identification method and the system of a kind of gate channel system, gate passage.The employing embodiment of the invention can be by classification, number, direction of motion, movement velocity and attaching relation simple, that accurate, reliable method identifies the current target in the gate passage.
At first, the embodiment of the invention provides a kind of current target identification method of gate passage, and this method comprises:
Gather the output state of the gait sensor in the described gate passage;
Behind the output state that collects gait sensor output " disconnection ", in conjunction with front and back frame difference method described gate passage is carried out video acquisition by the background subtraction point-score;
The video information that collects according to background subtraction point-score and front and back frame difference method, and in conjunction with the contour area of current target in the described gate passage of output state information acquisition of the gait sensor that collects;
Classification, number by the current target that comprises in conjunction with the described contour area of output state information Recognition of gait sensor based on the AdaBoost sorting algorithm of Haar-Like feature;
According to the video information of continuous acquisition and the output state information of gait sensor the current target in the described contour area is followed the tracks of, obtain position, direction of motion, movement velocity and the attaching relation of described current target.
Accordingly, the embodiment of the invention also provides a kind of current target identification system of gate passage, and this system comprises:
The gait sensor;
The gait collecting unit is connected with described gait sensor, is used to gather the output state of described gait sensor;
Video acquisition unit is connected with described gait collecting unit, is used for after described gait collecting unit collects the output state of gait sensor output " disconnection " described gate passage being carried out video acquisition;
Profile obtains the unit, is used for the video information that collects according to described video acquisition unit, and the contour area of current target in the described gate passage of output state information acquisition of the gait sensor that collects in conjunction with described gait collecting unit;
The Classification and Identification unit is used for classification, number by the current target that comprises in conjunction with the described contour area of output state information Recognition of gait sensor based on the AdaBoost sorting algorithm of Haar-Like feature;
Tracking cell is used for according to the video information of continuous acquisition and the output state information of gait sensor the current target of described contour area being followed the tracks of, and obtains position, direction of motion, movement velocity and the attaching relation of described current target.
Accordingly, the embodiment of the invention also provides a kind of gate channel system, and this system includes the current target identification system of body, barrier door, ticket checking system and above-mentioned gate passage.
The embodiment of the invention by gathering the gait sensor in the gate passage output state information and the video information of gate passage collection discerned classification, number, direction of motion, movement velocity and the attaching relation of the current target in the described gate passage, realized by simple, accurately, method identifies the purpose of classification, number, direction of motion, movement velocity and the attaching relation of the current target in the gate passage reliably.
Description of drawings
Fig. 1 is that an example structure of the current target identification system of a kind of gate passage of the present invention is formed synoptic diagram;
Fig. 2 is an embodiment schematic flow sheet of the logical current target identification method of a kind of gate of the present invention;
Fig. 3 is the installation outboard profile of an embodiment of gate channel system of the present invention.
Embodiment
For making the purpose, technical solutions and advantages of the present invention clearer, the present invention is described in further detail below in conjunction with the accompanying drawings and the specific embodiments.
A kind of gate channel system that the embodiment of the invention provides includes the current target identification system of body, barrier door, ticket checking system and above-mentioned gate passage.Wherein, described body, barrier door and ticket checking system are the prior art content, and at this repeated description not, Fig. 1 is that an example structure of the current target identification system of a kind of gate passage of the present invention is formed synoptic diagram; As shown in Figure 1, the current target identification system of the gate passage of present embodiment comprises gait sensor 1, gait collecting unit 2, video acquisition unit 3 and CPU (central processing unit) 4, wherein,
Described gait sensor 1 comprises the photocell pickoff of a plurality of correlation, described a plurality of correlation sensor comprises many emission sensor and receiving sensors to mutual pairing, when passing through target (in the specific implementation, may single people, the people carries situations such as goods, many people be adjacent) in the gate passage when current, the receiving sensor that the light that emission sensor in the described correlation sensor is sent can be matched receives, at this moment, the output state of receiving sensor output " conducting "; Current in the gate passage as current target, make the light that sends of the part emission sensor in the described correlation sensor be blocked, at this moment, can not receive the light that corresponding emission sensor is sent with the receiving sensor of described emission sensor pairing, described receiving sensor will be exported the output state of " disconnection " as this output to sensor.In the present embodiment, described a plurality of correlation sensor with emission sensor and the receiving sensor the latter half of arranging the left and right sides sidewall that is installed in described gate passage respectively in the horizontal linear mode, general straight line equal intervals (spacing is smaller or equal to 18cm) parallel to the ground between normal adult knee and ankle, when pedestrian or animal were current in the gate passage, the output state of gait sensor can reflect the gait of described pedestrian or animal like this.
Described gait collecting unit 2 is connected with described gait sensor 1, is used to gather the output state of described gait sensor 1.In the specific implementation, gait collecting unit 2 with the sample frequency set (such as, 100HZ) the output state of the described gait sensor 1 of collection, such as, if the sensor in the passage is 8 pairs, at the first sampling instant t1, a group traveling together has just entered the gate passage, the 1st of correlation sensor, the light that emission sensor in the 2 pairs of sensors is sent is blocked, cause to receive the light that corresponding emission sensor is sent with the receiving sensor of its pairing, the described the 1st, the output state of the receiving sensor output " disconnections " in the 3 pairs of sensors (suppose to represent disconnection) with logical symbol " 0 ", the light that other sensor sends in the passage can not be blocked, therefore, other receiving sensor will be exported the output state (supposing with logical symbol " 1 " expression conducting) of " conducting " in the passage, like this t1 constantly the output state of the gait sensor that collects of gait collecting unit 2 be " 01011111 "; At the second sampling instant t2, described pedestrian moves in the gate passage, the 5th of correlation sensor, the light that emission sensor in the 7 pairs of sensors is sent is blocked, cause to receive the light that corresponding emission sensor is sent with the receiving sensor of its pairing, the described the 5th, the output state of the receiving sensor output " disconnections " in the 7 pairs of sensors (suppose to represent disconnection) with logical symbol " 0 ", the light that other sensor sends in the passage can not be blocked, therefore, other receiving sensor will be exported the output state (supposing with logical symbol " 1 " expression conducting) of " conducting " in the passage, like this t2 constantly the output state of the gait sensor that collects of gait collecting unit 2 be " 11110101 ".
Described video acquisition unit 3 is connected with described gait collecting unit 2, is used for after described gait collecting unit 2 collects the output state of gait sensor 1 output " disconnection " described gate passage being carried out video acquisition.Described video acquisition unit 2 can be carried out video acquisition in conjunction with front and back frame difference method to described gate passage by the background subtraction point-score; Suppose that background subtraction partial image function is F
Bd(s, t), the target image of this function representation moment t and the difference of background image.Front and back frame difference image function is F
Td(s, t), the difference of front and back two field picture during this function representation moment t.T1, T2, T3 are for distinguishing three threshold values that target image and background image are provided with, supposing video acquisition unit 3 confirms by above-mentioned three threshold values whether target pixel point is the picture element of motion, such as supposing to work as F
Bd(s, t)＜T
1And F
Td(s, t)＜T
2Think that then target pixel point s is indeclinable background dot, works as F
Td(s, t)＞T
3, this target pixel point point that is motion change then.Video acquisition unit 3 can collect corresponding video information by above-mentioned background image difference point-score in conjunction with front and back frame difference method like this.In the specific implementation, video acquisition unit 2 can be camera.
Described CPU (central processing unit) 4 is connected with described video acquisition unit 3 with described gait collecting unit 2 respectively, is used for classification, number, direction of motion, movement velocity and attaching relation that the output state information of the gait sensor 1 that collects according to described gait collecting unit 2 and video information that described video acquisition unit 3 collects are discerned the current target of described gate passage.Still with reference to figure 1, the CPU (central processing unit) 4 of present embodiment comprises that specifically profile obtains unit 41, Classification and Identification unit 42 and tracking cell 43, wherein,
Described profile obtains unit 41 and is used for the video information that collects according to described video acquisition unit 3, and the contour area of current target in the described gate passage of output state information acquisition of the gait sensor 1 that collects in conjunction with described gait collecting unit 2; Still with reference to figure 1, the profile of present embodiment obtains unit 41 and specifically comprises amending unit 410 and processing unit 411, wherein, described amending unit 410 is used for the video information that the described video acquisition unit 3 of output state information correction of the gait sensor 1 that collects according to described gait collecting unit 2 collects, and makes the described video information that collects consistent with the output state information of the described gait sensor that collects; Such as, when video acquisition unit 3 was carried out video acquisition in conjunction with front and back frame difference method to described gate passage by the background subtraction point-score, amending unit 410 was further revised the video information that described video acquisition unit 3 is gathered according to the output state of the gait sensor 1 that collects.Such as, there is current target the position a-quadrant of video information in passage that video acquisition unit 3 is gathered, and gait collecting unit 2 collects the output state of the sensor of relevant position and is " conducting ", and it is wrong to think that then video information detects, in this case, self-adaptation is revised T
1With T
2Value, make the described video information that collects consistent with the output state information of the described gait sensor that collects, prepare for the information of the information of later gait sensor and human body contour outline sensor cooperatively interacts.Processing unit 411 is used for that described amending unit 410 revised video informations are carried out connected domain to be handled, and obtains the contour area of current target in the described gate passage.In the specific implementation, detected motion pixel point formation prospect picture element.The prospect picture element is done the morphologic open and close computing of mathematics, remove isolated erroneous point, and connect correct target travel picture element, make the target travel picture element form connected region, obtain the contour area of current target in the described gate passage.
Described Classification and Identification unit 42 is used for classification, the number by the current target that comprises in conjunction with the described contour area of output state information Recognition of gait sensor based on the AdaBoost sorting algorithm of Haar-Like feature;
Still with reference to figure 1, the Classification and Identification unit 42 of present embodiment specifically comprises division unit 420, extraction unit 421 and recognition unit 422, wherein, division unit 420 is used for the output state of the gait sensor that collects according to gait collecting unit 2 and the scope of the contour area that described processing unit 411 obtains is divided into a plurality of independently contour areas with described contour area; In the specific implementation, division unit 420 can be divided into a plurality of independently contour areas with big contour area according to ordinary people's profile scope.Extraction unit 421 is used to extract the Haar-Like feature of described a plurality of independent contour areas; The Haar-Like feature comprises one or more in edge feature, linear feature and the broken line feature.Recognition unit 423 carries out analyzing and processing by the AdaBoost sorting algorithm to the Haar-Like feature of described extraction, and discerns the classification and the number of the current target that described a plurality of independent contour area comprises in conjunction with the output state of the gait sensor that collects.Recognition unit 423 has the principle of different Haar-Like features according to the current target of difference, by the AdaBoost sorting algorithm Haar-Like feature of described extraction is carried out analyzing and processing, and the principle of the different gaits that present in conjunction with different current targets is discerned the classification and the number of the current target that described a plurality of independent contour area comprises.In the specific implementation, each Haar-Like feature is made up of black and white two parts, and its feature calculation is as follows:
Haar_Like_Feauture＝w_b*sum(area_b)+w_w*sum(area_w)
The weight of w_b black region wherein, the weight of w_w white portion, sum (area_b) black part branch enclose image-region pixel and, the pixel of sum (area_w) image-region that white portion encloses with.
AdaBoost sorting algorithm based on the Haar-Like feature is implemented as follows:
AdaBoost integrates Weak Classifier, constitutes strong classifier, reaches the effect of classification then by the cascade of strong classifier, at first is defined as follows notion:
Given one group of training sample { (x
i, y
i), i=1 ..., N}, wherein y
i=0, and 1}, wherein m contains the image of human body contour outline, n image that does not have human body contour outline in the training sample.M+n=N wherein.If x
iBe the image that contains human body contour outline, then y
i=0, otherwise y
i=1.
The simple classification device form of j Haar-Like feature generation is:
Wherein: h
jThe value of expression simple classification device; θ
jBe threshold value; p
jThe direction of the expression sign of inequality, can only get ± 1; f
j(x) representation feature value.
Concrete steps are as follows:
1) initialization error weight is for y
i=0 sample, ω
1, i=1/2m; For y
i=1 sample,
ω
1，i＝1/2n。
2) to each t=1 ..., T (wherein T is the number of times of training):
2.1) weight normalization,
2.2) for each feature j, generate corresponding simple classification device h according to top method
j, calculate error with respect to current weight:
2.3) select to have least error ε
tSimple classification device h
tJoin in the strong classifier and go.
2.4) upgrade the pairing weight of each sample:
If i sample x
iCorrectly classified, then e
i=0, on the contrary e
i=1,
3) strong classifier that forms at last is:
Wherein
At last strong classifier is together in series, forms the classification device, reach and distinguish current other purpose of target class.
Described tracking cell 43 is used for according to the output state information of the video information of continuous acquisition and gait sensor the current target of described contour area being followed the tracks of, and obtains position, direction of motion, movement velocity and the attaching relation of described current target.In the specific implementation, video acquisition unit 3 is gathered video information with the sample frequency of setting (such as 25 frame/seconds), and tracking cell 43 carries out association with the video information of consecutive frame, and thinks that different frame the image identical or adjacent position is a same target; Same, gait collecting unit 2 is also gathered the output state of gait sensor with the sample frequency of setting (foregoing 100HZ), and tracking cell 43 carries out association with the gait sensor output state information of consecutive frame; Tracking cell 43 can obtain position, direction of motion, movement velocity and the attaching relation of current target in the gate passage according to the video information of continuous acquisition and gait sensor output state information like this.
Accordingly, the embodiment of the invention provides a kind of current target identification method of gate passage, and this method comprises: the output state of gathering the gait sensor in the described gate passage; Behind the output state that collects gait sensor output " disconnection ", described gate passage is carried out video acquisition; Discern classification, number, direction of motion, movement velocity and the attaching relation of the current target in the described gate passage according to the output state information of the gait sensor that collects and the video information that collects.
Concrete, Fig. 2 is an embodiment schematic flow sheet of the current target identification method of a kind of gate passage of the present invention.As shown in Figure 2, the current target identification method of the gate passage of present embodiment comprises:
Step S100 gathers the output state of the gait sensor in the described gate passage; Described gait sensor comprises the photocell pickoff of a plurality of correlation, described a plurality of correlation sensor comprises many emission sensor and receiving sensors to mutual pairing, when passing through target (in the specific implementation, may single people, the people carries situations such as goods, many people be adjacent) in the gate passage when current, the receiving sensor that the light that emission sensor in the described correlation sensor is sent can be matched receives, at this moment, the output state of receiving sensor output " conducting "; Current in the gate passage as current target, make the light that sends of the part emission sensor in the described correlation sensor be blocked, at this moment, can not receive the light that corresponding emission sensor is sent with the receiving sensor of described emission sensor pairing, described receiving sensor will be exported the output state of " disconnection " as this output to sensor.In the present embodiment, described a plurality of correlation sensor with emission sensor and the receiving sensor the latter half of arranging the left and right sides sidewall that is installed in described gate passage respectively in the horizontal linear mode, general straight line equal intervals (spacing is smaller or equal to 18cm) parallel to the ground between normal adult knee and ankle, when pedestrian or animal were current in the gate passage, the output state of gait sensor can reflect the gait of described pedestrian or animal like this.In the specific implementation, with the sample frequency set (such as, 100HZ) the output state of the described gait sensor of collection, such as, if the sensor in the passage is 8 pairs, at the first sampling instant t1, a group traveling together has just entered the gate passage, the 1st of correlation sensor, the light that emission sensor in the 2 pairs of sensors is sent is blocked, cause to receive the light that corresponding emission sensor is sent with the receiving sensor of its pairing, the described the 1st, the output state of the receiving sensor output " disconnections " in the 3 pairs of sensors (suppose to represent disconnection) with logical symbol " 0 ", the light that other sensor sends in the passage can not be blocked, therefore, other receiving sensor will be exported the output state (supposing with logical symbol " 1 " expression conducting) of " conducting " in the passage, and the output state of the gait sensor that collects constantly of t1 is " 01011111 " like this; At the second sampling instant t2, described pedestrian moves in the gate passage, the 5th of correlation sensor, the light that emission sensor in the 7 pairs of sensors is sent is blocked, cause to receive the light that corresponding emission sensor is sent with the receiving sensor of its pairing, the described the 5th, the output state of the receiving sensor output " disconnections " in the 7 pairs of sensors (suppose to represent disconnection) with logical symbol " 0 ", the light that other sensor sends in the passage can not be blocked, therefore, other receiving sensor will be exported the output state (supposing with logical symbol " 1 " expression conducting) of " conducting " in the passage, and the output state of the gait sensor that collects constantly of t2 is " 11110101 " like this.
Step S101 behind the output state that collects gait sensor output " disconnection ", carries out video acquisition in conjunction with front and back frame difference method to described gate passage by the background subtraction point-score; In the specific implementation, can carry out video acquisition in conjunction with front and back frame difference method to described gate passage by the background subtraction point-score; Suppose that background subtraction partial image function is F
Bd(s, t), the target image of this function representation moment t and the difference of background image.Front and back frame difference image function is F
Td(s, t), the difference of front and back two field picture during this function representation moment t.T1, T2, T3 are for distinguishing three threshold values that target image and background image are provided with, supposing video acquisition unit 3 confirms by above-mentioned three threshold values whether target pixel point is the picture element of motion, such as supposing to work as F
Bd(s, t)＜T
1And F
Td(s, t)＜T
2Think that then target pixel point s is indeclinable background dot, works as F
Td(s, t)＞T
3, this target pixel point point that is motion change then.Can collect corresponding video information by above-mentioned background image difference point-score in conjunction with front and back frame difference method like this.
Step S102, according to the output state information correction of the gait sensor that collects described according to the background subtraction point-score in conjunction with the described video information that front and back frame difference method collects, make the described video information that collects consistent with the output state information of the described gait sensor that collects; Such as, there is current target the position a-quadrant of the video information of collection in passage, and the output state of sensor that collects the relevant position is for " conducting ", it is wrong to think that then video information detects, and in this case, self-adaptation is revised T
1With T
2Value, make the described video information that collects consistent with the output state information of the described gait sensor that collects, prepare for the information of the information of later gait sensor and human body contour outline sensor cooperatively interacts.
Step S103 carries out connected domain to revised video information and handles, and obtains the contour area of current target in the described gate passage; In the specific implementation, detected motion pixel point formation prospect picture element.The prospect picture element is done the morphologic open and close computing of mathematics, remove isolated erroneous point, and connect correct target travel picture element, make the target travel picture element form connected region, obtain the contour area of current target in the described gate passage.
Step S104 is divided into a plurality of independently contour areas according to the output state of the gait sensor that collects and the scope of described contour area with described contour area; In the specific implementation, can big contour area be divided into a plurality of independently contour areas according to ordinary people's profile scope.
Step S105 extracts the Haar-Like feature of described a plurality of independently contour areas; The Haar-Like feature comprises one or more in edge feature, linear feature and the broken line feature.
Step S106 carries out analyzing and processing by the AdaBoost sorting algorithm to the Haar-Like feature of described extraction, and the classification and the number of the current target that comprises in conjunction with the described a plurality of independent contour areas of output state information Recognition of the gait sensor that collects.The principle that has different Haar-Like features according to the current target of difference, by the AdaBoost sorting algorithm Haar-Like feature of described extraction is carried out analyzing and processing, and the principle of the different gaits that present in conjunction with different current targets is discerned the classification and the number of the current target that described a plurality of independent contour area comprises.In the specific implementation, each Haar-Like feature is made up of black and white two parts, and its feature calculation is as follows:
Haar_Like_Feauture＝w_b*sum(area_b)+w_w*sum(area_w)
The weight of w_b black region wherein, the weight of w_w white portion, sum (area_b) black part branch enclose image-region pixel and, the pixel of sum (area_w) image-region that white portion encloses with.
AdaBoost sorting algorithm based on the Haar-Like feature is implemented as follows:
AdaBoost integrates Weak Classifier, constitutes strong classifier, reaches the effect of classification then by the cascade of strong classifier, at first is defined as follows notion:
Given one group of training sample { (x
i, y
i), i=1 ..., N}, wherein y
i=0, and 1}, wherein m contains the image of human body contour outline, n image that does not have human body contour outline in the training sample.M+n=N wherein.If x
iBe the image that contains human body contour outline, then y
i=0, otherwise y
i=1.
The simple classification device form of j Haar-Like feature generation is:
Wherein: h
jThe value of expression simple classification device; θ
jBe threshold value; p
jThe direction of the expression sign of inequality, can only get ± 1; f
j(x) representation feature value.
Concrete steps are as follows:
1) initialization error weight is for the sample of yx=0, ω
1, i=1/2m; For y
i=1 sample, ω
1, i=1/2n.
2) to each t=1 ..., T (wherein T is the number of times of training):
2.1) weight normalization,
2.2) for each feature j, generate corresponding simple classification device hj according to top method, calculate error with respect to current weight:
2.3) select to have least error ε
tSimple classification device h
tJoin in the strong classifier and go.
2.4) upgrade the pairing weight of each sample:
If i sample x
iCorrectly classified, then e
i=0, on the contrary e
i=1,
3) strong classifier that forms at last is:
Wherein
At last strong classifier is together in series, forms the classification device, reach and distinguish current other purpose of target class.
Step S107 follows the tracks of the current target in the described contour area according to the video information of continuous acquisition and the output state information of gait sensor, obtains position, direction of motion, movement velocity and the attaching relation of described current target.In the specific implementation, gather video information, the video information of consecutive frame is carried out association, and think that different frame the image identical or adjacent position is a same target with the sample frequency of setting (such as 25 frame/seconds); Same, also gather the output state of gait sensor with the sample frequency of setting (foregoing 100HZ), tracking cell 43 carries out association with the gait sensor output state information of consecutive frame; Can obtain position, direction of motion, movement velocity and the attaching relation of current target in the gate passage like this according to the video information of continuous acquisition and gait sensor output state information.
Fig. 3 is the installation outboard profile of an embodiment of gate channel system of the present invention.The body 201 of gate channel system one side of knowing clearly shown in Fig. 3, barrier door 203, and current target identification system in gait sensor 202 and the camera 200 in the current target identification system (in the specific implementation, other modular units of the current target identification system of the ticket checking system of gate channel system and passage are arranged on internal body, so these modules are not shown in Fig. 3), wherein, described gait sensor 202 is arranged the latter half of the body inner wall that is installed in described gate passage in the horizontal linear mode.The height on each relative ground of gait sensor is 15～45 centimetres among the figure, and the spacing between the adjacent gait sensor is not more than 18 centimetres.Described camera 200 is arranged on the top of described body 201, and the picked-up scope of camera 200 comprises barrier door prepass length 1/6 zone (there is shown three positions that camera can be provided with) at least.Present embodiment passes through classification, number, direction of motion, movement velocity and the attaching relation of the output state of collection gait sensor 202 in conjunction with the current target of the video information double-point information identification gate passage of camera 200 collections, has realized by purpose simple, that accurate, reliable method identifies classification, number, direction of motion, movement velocity and the attaching relation of the current target in the gate passage.
The embodiment of the invention by gathering the gait sensor in the gate passage output state information and the video information of gate passage collection discerned classification, number, direction of motion, movement velocity and the attaching relation of the current target in the described gate passage, realized by simple, accurately, method identifies the purpose of classification, number, direction of motion, movement velocity and the attaching relation of the current target in the gate passage reliably.
More than cited only be a kind of preferred embodiment of the present invention, can not limit the present invention's interest field certainly with this, therefore the equivalent variations of doing according to claim of the present invention still belongs to the scope that the present invention is contained.
Claims (12)
1. the current target identification method of a gate passage is characterized in that, comprising:
Gather the output state of the gait sensor in the described gate passage;
Behind the output state that collects gait sensor output " disconnection ", in conjunction with front and back frame difference method described gate passage is carried out video acquisition by the background subtraction point-score;
The video information that collects according to background subtraction point-score and front and back frame difference method, and in conjunction with the contour area of current target in the described gate passage of output state information acquisition of the gait sensor that collects;
Classification, number by the current target that comprises in conjunction with the described contour area of output state information Recognition of gait sensor based on the AdaBoost sorting algorithm of Haar-Like feature;
According to the video information of continuous acquisition and the output state information of gait sensor the current target in the described contour area is followed the tracks of, obtain position, direction of motion, movement velocity and the attaching relation of described current target.
2. the current target identification method of gate passage as claimed in claim 1, it is characterized in that, the described video information that collects according to background subtraction point-score and front and back frame difference method, and specifically comprise in conjunction with the step of the contour area of current target in the described gate passage of the output state information acquisition of gait sensor:
According to the output state information correction of the gait sensor that collects described according to the background subtraction point-score in conjunction with the described video information that front and back frame difference method collects, make the described video information that collects consistent with the output state information of the described gait sensor that collects;
Revised video information is carried out connected domain handle, obtain the contour area of current target in the described gate passage.
3. the current target identification method of gate passage as claimed in claim 1, it is characterized in that described classification by the current target that comprises in conjunction with the described contour area of output state information Recognition of gait sensor based on the AdaBoost sorting algorithm of Haar-Like feature, the step of number specifically comprise:
According to the output state of the gait sensor that collects and the scope of described contour area described contour area is divided into a plurality of independently contour areas;
Extract the Haar-Like feature of described a plurality of independently contour areas;
By the AdaBoost sorting algorithm Haar-Like feature of described extraction is carried out analyzing and processing, and the classification and the number of the current target that comprises in conjunction with the described a plurality of independent contour areas of output state information Recognition of the gait sensor that collects.
4. the current target identification method of gate passage as claimed in claim 3 is characterized in that, described Haar-Like feature comprises one or more in edge feature, linear feature and the broken line feature.
5. the current target identification system of a gate passage is characterized in that, comprising:
The gait sensor;
The gait collecting unit is connected with described gait sensor, is used to gather the output state of described gait sensor;
Video acquisition unit is connected with described gait collecting unit, is used for after described gait collecting unit collects the output state of gait sensor output " disconnection " described gate passage being carried out video acquisition;
Profile obtains the unit, is used for the video information that collects according to described video acquisition unit, and the contour area of current target in the described gate passage of output state information acquisition of the gait sensor that collects in conjunction with described gait collecting unit;
The Classification and Identification unit is used for classification, number by the current target that comprises in conjunction with the described contour area of output state information Recognition of gait sensor based on the AdaBoost sorting algorithm of Haar-Like feature;
Tracking cell is used for according to the video information of continuous acquisition and the output state information of gait sensor the current target of described contour area being followed the tracks of, and obtains position, direction of motion, movement velocity and the attaching relation of described current target.
6. the current target identification system of gate passage as claimed in claim 5 is characterized in that, described profile obtains the unit and specifically comprises:
Amending unit, be used for the video information that the described video acquisition unit of output state information correction of the gait sensor that collects according to described gait collecting unit collects, make the described video information that collects consistent with the output state information of the described gait sensor that collects;
Processing unit is used for that the revised video information of described amending unit is carried out connected domain and handles, and obtains the contour area of current target in the described gate passage.
7. the current target identification system of gate passage as claimed in claim 6 is characterized in that, described Classification and Identification unit specifically comprises:
Division unit is used for the output state of the gait sensor that collects according to the gait collecting unit and the scope of the contour area that described processing unit obtains described contour area is divided into a plurality of independently contour areas;
Extraction unit is used to extract the Haar-Like feature of described a plurality of independent contour areas;
Recognition unit carries out analyzing and processing by the AdaBoost sorting algorithm to the Haar-Like feature of described extraction, and discerns the classification and the number of the current target that described a plurality of independent contour area comprises in conjunction with the output state of the gait sensor that collects.
8. gate channel system comprises the current target identification system of body, barrier door, ticket checking system and the described gate passage of claim 5.
9. gate channel system as claimed in claim 8 is characterized in that, described gait sensor is arranged the latter half of the body inner wall that is installed in described gate passage in the horizontal linear mode.
10. gate channel system as claimed in claim 9 is characterized in that, the height on the relative ground of described gait sensor is 15～45 centimetres, and the spacing between the adjacent gait sensor is not more than 18 centimetres.
11. gate channel system as claimed in claim 8 is characterized in that described video acquisition unit comprises the video capture device, this video capture device is arranged on the top of described body.
12. gate channel system as claimed in claim 11 is characterized in that, the picked-up scope of described video capture device comprises barrier door prepass length 1/6 zone at least.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2007100313843A CN101436314B (en) | 2007-11-14 | 2007-11-14 | Brake machine channel system, method and system for recognizing passing object in brake machine channel |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2007100313843A CN101436314B (en) | 2007-11-14 | 2007-11-14 | Brake machine channel system, method and system for recognizing passing object in brake machine channel |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101436314A CN101436314A (en) | 2009-05-20 |
| CN101436314B true CN101436314B (en) | 2010-08-11 |
Family
ID=40710744
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2007100313843A Active CN101436314B (en) | 2007-11-14 | 2007-11-14 | Brake machine channel system, method and system for recognizing passing object in brake machine channel |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101436314B (en) |
Families Citing this family (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102024275B (en) * | 2010-12-21 | 2012-10-24 | 南京大学 | Automatic ticket checking device |
| CN106127866B (en) * | 2016-06-29 | 2019-03-29 | 北京明生宏达科技有限公司 | Ticket checking method and channel management equipment |
| CN106127865B (en) * | 2016-06-29 | 2019-03-29 | 北京明生宏达科技有限公司 | Ticket checking method and channel management equipment |
| CN106127901B (en) * | 2016-06-29 | 2019-08-13 | 北京明生宏达科技有限公司 | The current channel management equipment of left and right ticket checking and left and right ticket checking passing method |
| CN106127902B (en) * | 2016-06-29 | 2019-03-29 | 北京明生宏达科技有限公司 | The current channel management equipment of left and right ticket checking and left and right ticket checking passing method |
| CN107221056B (en) * | 2017-05-25 | 2019-10-08 | 深圳市捷成安科技有限公司 | The method stopped based on human bioequivalence |
| CN107221058A (en) * | 2017-05-25 | 2017-09-29 | 刘萍 | Intelligent channel barrier system |
| CN111192391B (en) * | 2018-10-25 | 2022-09-23 | 杭州海康威视数字技术股份有限公司 | Pedestrian passageway gate control method and device based on images and/or videos |
| CN110287778B (en) * | 2019-05-15 | 2021-09-10 | 北京旷视科技有限公司 | Image processing method and device, terminal and storage medium |
| CN110288632A (en) * | 2019-05-15 | 2019-09-27 | 北京旷视科技有限公司 | A kind of image processing method, device, terminal and storage medium |
| CN112927400A (en) * | 2021-02-02 | 2021-06-08 | 艾弗世(苏州)专用设备股份有限公司 | Visual passing logic gate device, visual passing logic method and passing method |
| CN113140093A (en) * | 2021-04-26 | 2021-07-20 | 贵阳职业技术学院 | Fatigue driving detection method based on AdaBoost algorithm |
| CN113420693B (en) * | 2021-06-30 | 2022-04-15 | 成都新潮传媒集团有限公司 | Door state detection method and device, and car passenger flow statistical method and equipment |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB2342208A (en) * | 1998-09-25 | 2000-04-05 | Fujitsu Ltd | System and method for customer recognition |
| JP2003009137A (en) * | 2001-06-26 | 2003-01-10 | Toshiba Electronic Systems Co Ltd | Supervising system |
| JP2005031955A (en) * | 2003-07-11 | 2005-02-03 | Kddi Corp | Mobile body tracking system |
| JP2006163788A (en) * | 2004-12-07 | 2006-06-22 | Hitachi Communication Technologies Ltd | Visitor management method and management system |
| CN1932888A (en) * | 2006-09-28 | 2007-03-21 | 广州市地下铁道总公司 | Pedestrian passing identifying method and apparatus for gate type brake machine |
| CN101017573A (en) * | 2007-02-09 | 2007-08-15 | 南京大学 | Method for detecting and identifying moving target based on video monitoring |
-
2007
- 2007-11-14 CN CN2007100313843A patent/CN101436314B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB2342208A (en) * | 1998-09-25 | 2000-04-05 | Fujitsu Ltd | System and method for customer recognition |
| JP2003009137A (en) * | 2001-06-26 | 2003-01-10 | Toshiba Electronic Systems Co Ltd | Supervising system |
| JP2005031955A (en) * | 2003-07-11 | 2005-02-03 | Kddi Corp | Mobile body tracking system |
| JP2006163788A (en) * | 2004-12-07 | 2006-06-22 | Hitachi Communication Technologies Ltd | Visitor management method and management system |
| CN1932888A (en) * | 2006-09-28 | 2007-03-21 | 广州市地下铁道总公司 | Pedestrian passing identifying method and apparatus for gate type brake machine |
| CN101017573A (en) * | 2007-02-09 | 2007-08-15 | 南京大学 | Method for detecting and identifying moving target based on video monitoring |
Also Published As
| Publication number | Publication date |
|---|---|
| CN101436314A (en) | 2009-05-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101436314B (en) | Brake machine channel system, method and system for recognizing passing object in brake machine channel | |
| CN109522793B (en) | Method for detecting and identifying abnormal behaviors of multiple persons based on machine vision | |
| Liu et al. | A fall detection system using k-nearest neighbor classifier | |
| CN101980245B (en) | Adaptive template matching-based passenger flow statistical method | |
| US20100202657A1 (en) | System and method for object detection from a moving platform | |
| CN102903122B (en) | Video object tracking method based on feature optical flow and online ensemble learning | |
| CN110378179B (en) | Subway ticket evasion behavior detection method and system based on infrared thermal imaging | |
| CN103366154B (en) | Reconfigurable clear path detection system | |
| EP3576008A1 (en) | Image based lane marking classification | |
| CN104902258A (en) | Multi-scene pedestrian volume counting method and system based on stereoscopic vision and binocular camera | |
| CN109145742A (en) | A kind of pedestrian recognition method and system | |
| CN102419819A (en) | Method and system for recognizing human face image | |
| CN104881637A (en) | Multimode information system based on sensing information and target tracking and fusion method thereof | |
| CN109472226B (en) | Sleeping behavior detection method based on deep learning | |
| EP2709066A1 (en) | Concept for detecting a motion of a moving object | |
| CN111259718B (en) | Staircase retention detection method and system based on Gaussian mixture model | |
| CN107315993A (en) | A kind of peephole system and its face identification method based on recognition of face | |
| CN104933424A (en) | Vehicle and pedestrian monitoring method and apparatus | |
| CN104657724A (en) | Method for detecting pedestrians in traffic videos | |
| CN104794451A (en) | Human identification method based on block matching structure | |
| CN103106414A (en) | Detecting method of passer-bys in intelligent video surveillance | |
| CN101739685B (en) | Moving object classification method and system | |
| CN112836683A (en) | License plate recognition method, device, equipment and medium for portable camera equipment | |
| CN115100716A (en) | Intelligent community pedestrian tracking and positioning method and system | |
| CN101908150A (en) | Human body detection method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |