KR101786734B1 - Apparatus and method for thinning characters - Google Patents

Abstract

The present invention relates to a method and apparatus for recognizing a character existing in an image input from a scanner or a camera or an image captured in a digital screen, The pixel corresponding to the characteristic portion of each character is set as a main pixel in advance so as not to be deleted through the image processing of the thinning process. Thus, even if the thinning process is performed, the original character before the thinning process can be accurately recognized .

Description

[0001] APPARATUS AND METHOD FOR THINNING CHARACTERS [0002]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to character recognition, and more particularly to recognition of characters existing in an image input from a scanner or a camera or an image captured by a digital screen In the thinning process of characters to reduce the variable due to the thickness variation of characters during the preprocessing and feature extraction process of character recognition, the image processing of the thinning process is performed on the pixels corresponding to characteristic portions of each character The present invention relates to a character thinning apparatus and method for accurately recognizing an original character before a thinning process even if the thinning process is performed.

2. Description of the Related Art In general, a scanner and a camera are most commonly used for inputting character images for character recognition, and also images are extracted directly from a computer monitor or a mobile terminal display device . We use a thinning method as preprocessing to separate characters from the input image and recognize the characters.

This thinning method is one of the preprocessing methods used in the field of character recognition and image processing, and is a technique for eliminating the thickness of a character or image and processing the skeleton to be left. The Zhang-Suen algorithm has been widely used as a simple and effective method for the thinning method. In this method, two partial iterations are performed, and black pixels corresponding to the condition are converted into white pixels until no more images change.

FIG. 1 illustrates a thinning operation control flow of a conventional character image. Referring to FIG. 1, the thinning operation will be described below.

First, when a character image scanned or photographed from a scanner or a camera is input (S100), a thinning process is performed on the pixel region of each character in the character image.

That is, it is checked whether there is one or more black pixels among the pixels neighboring the current black pixel with respect to each of the black pixels forming the character (S102). If there are one or more black pixels, It is checked whether the number of black pixels is 2 or more and 6 or less (S104).

In this case, when the number of black pixels adjacent to the periphery of the current black pixel is 2 or more and 6 or less, as shown in FIG. 3, If at least one of the pixels is a white pixel (S106), the current black pixel is converted into a white pixel to perform thinning of the character (S108).

Then, the thinning process is performed on all the black pixels in the character, and it is checked whether there is no white pixel change in the character image process (S110). If the thinning process is completed for all the black pixels, the thinning process is terminated do.

However, although the Zhang-Suen thinning method as shown in FIG. 1 as described above is used in various image processing related applications, it has some problems in application to character recognition.

FIG. 2 shows examples of the results obtained when the conventional thinning method is applied to an input character image. That is, in FIGS. 2 (a) and 2 (b), the letter "cow" shows that an important part of the letter disappears through the existing thinning process described above, and the letter "c" (D), the important part of the letter has disappeared after the thinning, and the letter "e" of the letter "e" is the letter " After stroke, important strokes have disappeared and have been replaced with other characters called 'pole'.

In addition, in the case of a character similar to the above example or a character of another type, the conventional method according to the image of the character image results in the above-described result after the thinning process. Therefore, even if a feature is extracted for character recognition with such a result, there is a limit to improve the recognition performance even if there is a character image in which important features of the character have disappeared by any good recognition methodology.

Accordingly, in recognizing characters existing in an image input from a scanner, a camera, or an image captured on a digital screen, it is possible to reduce the number of characters The pixel corresponding to the characteristic portion of each character in the line drawing process is set as the main pixel in advance so as not to be deleted through the image processing of the thinning process so that even if the thinning process is performed, And a character thinning apparatus and method for enabling a character thinning process.

The present invention relates to a character thinning device, comprising: a character image input unit for inputting a character image; a binarizing unit for converting a character image input from the character image input unit into binary data; And a thinning unit for thinning the character image while classifying the predetermined number of black pixels corresponding to the end portion of each character in the pixel region into main pixels so that the main pixels are not thinned do.

The thinning unit may check whether the number of black pixels neighboring the black pixel is included in a reference value range for a black pixel existing in a pixel region of each character, and if the number of neighboring black pixels is greater than the reference value The black pixel is converted into a white pixel when one or more white pixels among the pixels surrounding the black pixel in a cross shape are present.

Also, the thinning unit may check whether the number of black pixels neighboring the main pixel is included in the reference value range with respect to the main pixel, and when the number of neighboring black pixels is included in the reference value range, In the L-shaped shape are all white pixels, and if the pixels are all white pixels, the main pixels are not converted into white pixels.

Further, the reference value range is characterized in that the number of the black pixels is 2 or more and 6 or less.

The thinning unit may be configured to convert the main pixel into a white pixel if all four consecutive pixels surrounding the main pixel in an L shape are white pixels, and if not all of the pixels are white pixels.

According to another aspect of the present invention, there is provided a method of thinning a character image, the method comprising: inputting a character image; converting the character image into binary data; Setting a predetermined number of black pixels corresponding to the end portion as main pixels; and performing thinning so that the main pixels are not removed for the black pixels of each character.

The step of performing the thinning may include checking whether the number of black pixels neighboring the black pixel is included in a reference value range for a black pixel existing in a pixel region of each character, The method comprising the steps of: if the white pixel is included in the range of the reference value, checking whether at least one white pixel among the pixels surrounding the black pixel in a cross shape is included; .

The step of performing the thinning may include: checking whether the number of black pixels neighboring the main pixel is included in a reference value range with respect to the main pixel; determining whether the number of neighboring black pixels is within the reference value range The method comprising the steps of: determining whether all four consecutive pixels surrounding the main pixel in an L-shape are white pixels if all four consecutive pixels are white pixels; The method comprising the steps of:

Further, the reference value range is characterized in that the number of the black pixels is 2 or more and 6 or less.

The method may further include the step of converting the main pixel into a white pixel if the four consecutive pixels are not all white pixels.

In the present invention, in recognition of a character existing in an image input from a scanner or a camera or an image captured in a digital screen, a character thinning process for reducing a variable due to a change in thickness of characters during a process of pre- The pixel corresponding to the characteristic portion of each character is set as a main pixel in advance so as not to be deleted through the image processing of the thinning process. Thus, even if the thinning process is performed, the original character before the thinning process can be accurately recognized There is an advantage.

FIG. 1 is a flow chart of operation control for a thinning process of a conventional character image,
FIG. 2 is an illustration of an erroneous character variation in the conventional thinning process,
FIG. 3 is an exemplary illustration of a pixel check condition in the thinning process,
FIG. 4 is a diagram illustrating a main pixel setting in a character image according to an exemplary embodiment of the present invention,
5 is a diagram illustrating an example of a pixel check condition in a thinning process according to an exemplary embodiment of the present invention,
6 is a block diagram of a character image thinning device according to an embodiment of the present invention;
FIG. 7 is a flow chart of operation control for a thinning process of a character image according to an embodiment of the present invention;
8 is a diagram illustrating an example of a thinned character image through a thinning process according to an embodiment of the present invention.

Hereinafter, the operation principle of the present invention will be described in detail with reference to the accompanying drawings. In the following description of the present invention, detailed description of known functions and configurations incorporated herein will be omitted when it may make the subject matter of the present invention rather unclear. The following terms are defined in consideration of the functions of the present invention, and these may be changed according to the intention of the user, the operator, or the like. Therefore, the definition should be based on the contents throughout this specification.

First, in the conventional thinning method, when a black pixel appears in a certain pattern, when a main pixel starts to be removed, the black pixel disappears continuously, and the main part is mostly lost, As a result, as shown in Fig. 2, the phenomenon that the main part of the character is removed occurs as described above.

Accordingly, in order to prevent the black pixels of the main part from being removed in the example of the Korean character image according to the present invention, as shown in FIG. 4, the main pixels which should not be removed from the actual Hangul characters are set, By not being removed from the thinning process, even if the thinning process is performed, the original character before the thinning process can be accurately recognized.

Fig. 5 shows the condition of this main pixel discrimination. Referring to FIG. 5, in the thinning process of the character image according to the present invention, the gray regions in FIGS. 5A, 5B and 5C are all white pixels for pixels set as main pixels, No matter what value it is, the black area must be a black pixel. If these three conditions are satisfied, it is determined to be a main pixel, and it is not removed during the thinning process.

6 is a detailed block diagram of a character thinning device 600 according to an embodiment of the present invention. The character thinning device 600 of the present invention includes a character image input unit 602, a binarization unit 604, And a fine line section 606.

First, the character image input unit 602 receives a character image scanned from a scanner or photographed by a camera, and transmits the character image to the binarization unit 604. The binarization unit 604 converts the character image transmitted from the character image input unit 602 into binary data.

The thinning unit 606 classifies a predetermined number of pixels corresponding to the end portion of each character in the pixel region of each character into main pixels with respect to the binarized character image from the binarization unit 604, And performs thinning of the character image while not performing it.

That is, for the black pixels existing in the pixel region of each character, the thinning unit 606 checks whether the number of black pixels neighboring the black pixel is 2 or more and 6 or less, and if the number of neighboring black pixels is 2 or more and 6 or less , The black pixel is converted into a white pixel when at least one white pixel among the pixels surrounding the black pixel in a cross shape is present as in the condition shown in FIG.

The fine line section 606 checks whether the number of black pixels neighboring the main pixel is greater than or equal to 2 and less than or equal to 6 for the main pixel. If the number of adjacent black pixels is equal to or greater than 2 and equal to or less than 6, The main pixels are not converted into white pixels in the case of all white pixels by checking whether all four consecutive pixels surrounding the main pixels in an L shape are white pixels as in FIG.

FIG. 7 is a flowchart illustrating a thinning operation of a character image according to an exemplary embodiment of the present invention. Referring to FIG. 1, the thinning operation will be described below.

First, when a character image scanned or photographed from a scanner or a camera is input, a thinning process is performed on a pixel region of each character in the character image.

That is, a character image scanned or photographed from a scanner or a camera is input to the character image input unit 602. The character image input as described above is generated as binary data for pixels of each character in the character image by the binarization unit 604 And input to the thinning unit 606.

As described above, when the pixel data of each character in the character image is input (S700), the thinning unit 606 performs thinning for each character in the character image.

That is, the neo-rendering unit 606 first checks whether there is one or more black pixels among the pixels adjacent to the current black pixel for each of the black pixels forming each character (S702). If there are one or more black pixels, It is checked whether the number of black pixels existing around the current black pixel is 2 or more and 6 or less (S704).

In this case, checking whether the number of black pixels is within the range of 2 to 6, which is a predetermined reference value range, is intended to reduce a range of inspection for converting a black pixel to a white pixel. When the number of black pixels existing on the circumference is smaller than 2 The number of pixels having a number of black pixels greater than 6 is 7 or 8. This means that the current black pixel is almost surrounded by another black pixel, It is judged not to check whether or not the black pixel is converted.

Next, when the number of black pixels existing around the current black pixel is 2 or more and 6 or less, as shown in FIG. 3, It is checked whether at least one of the pixels is a white pixel (S706).

At this time, when at least one of the four pixels in the cross-hatched area of the pixels surrounding the current black pixel is a white pixel, conventionally, the current black pixel is converted into a white pixel, The inspection is again performed by applying the condition shown in FIG. 5 to the current black pixel.

That is, in the case where at least one of the four pixels in the area indicated by the cross-hatched gray among the pixels surrounding the current black pixel is a white pixel, the fine line section 606 again refers to the current black pixel as shown in FIG. 5 Likewise, it is checked whether all four consecutive pixels surrounding the current black pixel in an L shape are white pixels (S708).

At this time, if all four consecutive pixels surrounding the current black pixel in an L shape are white pixels, the thinning unit 606 does not convert the current black pixel into a white pixel (S710).

However, if all four consecutive pixels surrounding the current black pixel in an L shape are white pixels, the thinning unit 606 converts the current black pixel into a white pixel (S712).

Next, the thinning unit 606 performs the thinning process for the current black pixel as described above, and then performs the thinning process for all the black pixels while repeating the step S702 (S712) again, (S714), the thinning process is terminated.

7, the black pixel removal test for the character image is performed in two stages, and it is completed when none of the black pixels are removed from all the results of the second stage.

Hereinafter, the process of FIG. 7 will be summarized again. If a binary character image comes in as input, it is checked whether all black pixels of the character image input in the first step are removed or not, If it is satisfied, the corresponding black pixel is changed to a white pixel. The conditions are as follows.

First, it is checked whether there is one black area connected to the current black pixel. Secondly, it is checked whether there are two or more black pixels among eight pixels adjacent to the current black pixel. Thirdly, it is checked whether one of the gray regions shown in (a) of FIG. 3 is a white pixel. Lastly, it is checked whether the gray areas shown in FIG. 5 are all white pixels.

Next, in the second step of the second step, the condition for determining whether or not to remove the black pixels of the input character image is checked. If the condition is satisfied, the corresponding black pixel is changed to a white pixel. The conditions are as follows.

First, it is checked whether there is one black area connected to the current black pixel. Secondly, it is checked whether there are two or more black pixels among eight pixels adjacent to the current black pixel. Thirdly, it is checked whether one of the gray regions shown in (b) of FIG. 3 is a white pixel. Lastly, it is checked whether the gray areas shown in FIG. 5 are all white pixels.

From the first to the third condition, the same condition is checked in the conventional technology, and a fourth condition is added to prevent the main pixel from being removed. If the condition is satisfied through the above two-step test, black pixels are removed and the thinning process proceeds.

FIG. 8 shows a result of thinning an improved character image according to the thinning method according to an embodiment of the present invention. Referring to FIG. 8, it can be seen that, unlike the result of FIG. 2, which is a result of applying the conventional technique, the main part of the character is not removed and thinning is performed. Therefore, it is possible to prevent characters from being removed due to the improved thinning result, thereby reducing the possibility of confusion between characters in the character recognition method, thereby improving the overall character recognition performance.

As described above, according to the present invention, in recognition of a character existing in an image input from a scanner or a camera or an image captured in a digital screen, in the process of pre-processing character recognition and extracting features, The pixel corresponding to the characteristic portion of each character in the thinning process of the character is set as the main pixel in advance so as not to be deleted through the image processing of the thinning process so that even if the thinning process is performed, To be accurately recognized.

While the invention has been shown and described with reference to certain preferred embodiments thereof, it will be understood by those skilled in the art that various changes and modifications may be made without departing from the spirit and scope of the invention. Accordingly, the scope of the invention should not be limited by the described embodiments but should be defined by the appended claims.

In the embodiment of the present invention, in recognizing a character, in the process of thinning a character to reduce a variable due to a change in the thickness of the character during the preprocessing and feature extraction process of character recognition, even if the thinning process is performed, The character can be correctly recognized and used in various character processing apparatuses.

602: character image input unit 604: binarization unit
606:

Claims (10)

A character image input unit for inputting character images;
A binarization unit for converting the character image input from the character image input unit into binary data;
The method according to claim 1, wherein the binarized character image is subjected to thinning with respect to the pixel region of each character, and whether or not consecutive pixels surrounding the black pixel in an L-shape for all the black pixels existing in the pixel region of each character are white pixels And if it is a white pixel, classifies the black pixel as a main pixel and does not convert the main pixel into a white pixel,
The character thinning device comprising: [Claim 2 is abandoned upon payment of the registration fee.] The method according to claim 1,
Wherein the thinning-
The black pixels existing in the pixel region of each character are included in the reference value range and the number of black pixels neighboring the black pixel is included in the reference value range, The number of neighboring black pixels is included in the reference value range, and when one or more white pixels among the pixels at the position surrounding the cross shape are present, Into a white pixel. delete [Claim 4 is abandoned upon payment of the registration fee.] 3. The method of claim 2,
The reference value range may be,
Wherein the number of neighboring black pixels is 2 or more and 6 or less. [Claim 5 is abandoned upon payment of registration fee.] 3. The method of claim 2,
Wherein the thinning-
And converts the black pixel into a white pixel when all four consecutive pixels surrounding the black pixel in an L shape are not white pixels. Inputting character images;
Converting the character image into binary data; And
Performing thinning on a pixel region of each character in the binarized character image;
/ RTI >
Wherein the step of performing thinning comprises:
The black pixels existing in the pixel region of each character are checked to determine whether all the consecutive pixels surrounding the black pixel in an L shape are white pixels. If all the pixels are white pixels, the black pixels are classified as main pixels, Wherein the thinning process is performed so that the main pixel is not removed by including the step of not converting the pixel into a white pixel. [7] has been abandoned due to the registration fee. The method according to claim 6,
Wherein the step of performing thinning comprises:
Checking whether the number of black pixels neighboring the black pixel included in a pixel region of each character is included in a reference value range; And
If the number of the black pixels is within the reference value range, inspecting whether at least one of white pixels surrounding the black pixels in a cross shape is present;
Wherein the number of neighboring black pixels is included within a reference value range and the black pixels are converted into white pixels when one or more white pixels among the pixels surrounding the cross shape are included. Thinning method. delete [Claim 9 is abandoned upon payment of registration fee.] 8. The method of claim 7,
The reference value range may be,
Wherein the number of neighboring black pixels is 2 or more and 6 or less. [Claim 10 is abandoned upon payment of the registration fee.] 8. The method of claim 7,
Wherein the step of performing thinning comprises:
And converting the black pixel into a white pixel when all four consecutive pixels surrounding the black pixel in an L shape are not white pixels.

Images