JP2510722B2 - How to distinguish uppercase and lowercase letters in English - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to a method for discriminating between uppercase and lowercase letters in English in a character recognition device.

[Conventional technology]

Conventionally, in distinguishing uppercase and lowercase letters, the height of a certain line of the target character is set as the height of the uppercase letter, and the threshold value is obtained from the ratio of the height of the lowercase letter set in advance to that height, It is known that the height of the target character is compared with this threshold value to discriminate between uppercase and lowercase letters.

[Problems to be Solved by the Invention]

However, in this method, the height of the line may be the height of the lowercase letter in the case of only lowercase letters, and in the conventional method, this is erroneously calculated and compared with the height of the uppercase letter. There is a problem that it is erroneously determined.

Therefore, an object of the present invention is to use the fact that uppercase letters and lowercase letters have different glyphs in the alphabet, and obtain the uppercase letter height and the lowercase letter height separately, and determine these two values and the size of the target character. By distinguishing uppercase letters from lowercase letters, the accuracy of distinguishing uppercase letters from lowercase letters is improved.

[Means for solving the problem]

At least the size of the target English character should be normalized, and the character recognition device that recognizes uppercase and lowercase letters using the same standard pattern should distinguish uppercase letters from lowercase letters. If the character type is different between lowercase and lowercase, or if the character type is the same, and if the character type is different, it is determined whether the character is uppercase or lowercase, and the character height of the target character type is summed separately for uppercase and lowercase. , If the character type is the same, perform the process of storing the height of that character for one character line, and then calculate the average height of uppercase letters and lowercase letters respectively from the integrated value of the character height of the character types with different glyphs. Determine the uppercase / lowercase distinction threshold, and then compare the height of each character for the same character type of the above-mentioned glyph with this threshold to determine whether it is uppercase or lowercase. That.

[Action]

From the recognition result of the target character, it is determined whether the target character has different upper and lower case glyphs, or the same glyph, and if the characters have different glyphs, it is determined whether the character is uppercase or lowercase. Character heights are summed separately for uppercase and lowercase letters, and if the character type is the same, the height of the character is memorized to obtain the recognition result for one line. Calculate the average height of lowercase letters,
A threshold for discriminating uppercase letters and lowercase letters is obtained from these two values, and the heights of characters having the same character shape are compared with this threshold value to discriminate uppercase letters and lowercase letters.

〔Example〕

FIG. 1 is a flowchart showing an embodiment of the present invention.

First, character image data is extracted by a known image processing method (see), and the target character is recognized by a known method (see). Next, it is determined whether the target character is an alphabetic character from the recognition result (see). If the character is an alphabetic character, whether the character is the same in uppercase and lowercase, such as "C (c)", or "A (a)" It is determined whether the letter shape is different between upper case and lower case (see), and if it is a different character, it is determined whether it is upper case or lower case (see). Add the ones multiplied by the values in the relative height table (see),
If it is a lowercase letter, add the sum of the lowercase letters and the character height multiplied by the value in the relative table of the character height to it (see).

An example of the relative table is shown in FIG. Ie uppercase
The table values for T1 are all "1", but for the table values for lowercase T2, there are some character heights that are similar to uppercase, such as b, h, l. It is set to "0.5" to balance it with other small letters.

On the other hand, if the character type is the same in uppercase and lowercase,
Holds the character height (see). The above steps (1) to (3) are repeated to obtain the recognition result of one line (see). 1
After recognizing the line, the average value of each case is calculated from the integrated value of uppercase and lowercase letters and the number of characters, and the optimum uppercase and lowercase discrimination threshold value is obtained from the intermediate value of these two values (see). And the capital letters that you saved in step,
Calls the character heights of lowercase letters with the same character type (reference), compares each with a threshold value, and determines uppercase letters and child characters (reference). This step is repeated until there are no stored characters (see).

〔The invention's effect〕

According to the present invention, it is possible to discriminate between uppercase letters and lowercase letters even in English letters having only lowercase letters, and there is an advantage that the discrimination accuracy can be improved.

[Brief description of drawings]

FIG. 1 is a flow chart showing an embodiment of the present invention,
The figure is a schematic diagram showing a relative table. Code Explanation T1 …… Upper case table, T2 …… Lower case table.

Claims (1)

(57) [Claims] 1. A character recognition device for normalizing at least the size of a target alphabetic character and recognizing a capital letter and a small letter by using a standard pattern having the same upper and lower case, based on the recognition result by the character recognizing device. , The upper and lower case of the target character type is determined by determining whether the target character is in uppercase and lowercase, and if the glyphs have different glyphs or the same glyph has the same glyph. While summing separately, if the character type of the glyph is the same, the process of storing the height of that character is performed for one character line, and then the average height of uppercase letters and lowercase letters from the cumulative value of the character height of the character types with different glyphs Is calculated to obtain the upper and lower case discrimination threshold, and then the height of each character of the same character type of the above-mentioned glyph is compared with this threshold to determine whether it is capital or small. Uppercase English, characterized in that to determine character, lowercase determination method.