JP2875285B2 - Image normalization method - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to an image normalization method for normalizing an image size.

[Conventional technology]

In devices or systems that read and recognize images of characters of various sizes, as preprocessing for character recognition,
It is necessary to normalize the size of the input image.

Basic techniques for such image normalization include thinning normalization and OR normalization.

As shown in FIG. 4, thinning-out normalization is a technique of generating a normalized image smaller in size than the input image by extracting pixels at regular intervals and discarding other pixels. OR normalization is a method of generating a normalized image smaller in size than the input image by taking a logical sum (OR) of pixels within a certain interval as shown in FIG.

However, when such a method is applied to the normalization of a character image, thinning-out normalization tends to cause the disappearance or "shading" of thin lines, and OR-normalization tends to cause thinning of white space. Are known, and this is the cause of misrecognition of characters. FIGS. 6 and 7 show examples of such crushing and “shading”.

In consideration of such a problem, a thin line segment is detected as a pre-process of thinning normalization, and a process of adding at least one pixel to the thin line portion is performed. Prevention methods (JP-A-60-126780)
A method to prevent the collapse of white space in characters by taking the logical product of the image obtained by OR-normalizing the input pixels as it is and the image obtained by OR-normalizing the input image by shifting one pixel in each of the horizontal and vertical directions. No. 60-11086 has also been devised.

[Problems to be solved by the invention]

However, when any of these methods is applied to an image larger than twice the normalized size, there is a problem that the processing becomes complicated and a sufficient effect cannot be obtained.

Therefore, an object of the present invention is to provide an image normalization method that can collectively handle images having various sizes and the like without causing “shading” or crushing of images.

[Means for solving the problem]

The feature of the image normalization method of the present invention is that a line segment thinning process is performed on an input image as a preprocess of a normalized image generation process (OR normalization process) by a logical sum operation which is a basic normalization method. It is to do.

In addition, the direction and the number of thinning processes are determined according to the properties of the input image.

[Action]

The weakness of the OR normalization, that is, the property that thin blanks are likely to be crushed, can be compensated for by the line segment thinning processing before the OR normalization processing. In addition, OR normalization has an advantage in that unlike a thinning-out normalization, "smearing" or disappearance of a thin line segment is unlikely to occur. Therefore, it is possible to obtain a normalized image free of image collapse and “shading”.

In addition, by simply changing the number and direction of the thinning process according to the size of the input image, etc., and simply optimizing the effect of the thinning process, normalization without crushing or "shading" can be performed on various images having different sizes. Images can be obtained.

〔Example〕

Hereinafter, an embodiment of the present invention in the character recognition system shown in FIG. 1 will be described.

A character extracting unit 2 cuts out image data for one character from the original image data input from an original image input unit (file, scanner, etc.) 1 and stores it in a character image buffer 4 in an image memory 3. Is passed to the normalization processing unit 5. This normalization processing unit 5 is a part directly related to the present invention.

In the normalization processing unit 5, the narrowing processing direction / number control unit 6 determines the direction and the number of times of performing the narrowing processing based on the passed information such as the size of the character, and designates the narrowing processing unit 7. Next, the thinning processing unit 7 executes the thinning processing of the line portion from the specified direction on the character image in the character image buffer 4 a specified number of times. The character image after the thinning process remains in the character image buffer 4. When the thinning process is completed, the OR processing unit 8 performs an OR normalization process on the character image read from the character image buffer 4 and stores the data of the generated normalized character image in the normalized image buffer 9 in the image memory 3. To be stored. The character recognition unit 10 performs a character recognition process on this image.

The thinning process will be described with reference to FIGS.
There are four possible narrowing directions: top to bottom, bottom to top, left to right, right to left.

FIG. 2 is an explanatory diagram of a narrowing process in a direction from top to bottom. Looking at the pixels from top to bottom, two black pixels appear in that direction.
In the part where more than one overlap, the top black pixel is 1
Replace with individual white pixels (eliminate one black pixel). In this example, the top black pixel in the portion marked with ↓ is erased, but the black pixel in the portion marked with ○ is not erased because there is only one black pixel.

FIG. 3 is an explanatory diagram of the narrowing process from the bottom to the top. In this case, at the portion where two or more black pixels overlap from the bottom to the top (the portion marked with ↑), one black pixel at the bottom is erased, but at the portion marked with ○, there is one black pixel. So leave without erasing.

By such a process, line segments do not break,
Horizontal line segments can be narrowed.

The same applies to the horizontal narrowing process, and a vertical line segment can be narrowed without generating a break in the line segment.

The direction of the thinning process is determined according to the character of the character image. For example, the scanner used to read the original image
If the character has a characteristic that makes it easy to erase horizontal line segments of a character, or if a horizontal line segment targets a thick font, only the vertical direction is selected. Conversely, in the case of a scanner in which the vertical lines of characters are easily crushed or in the case of fonts with thick vertical lines, thinning processing only in the horizontal direction is selected. Naturally, the upper, lower, left, and right directions may be selected.

By appropriately determining the direction of the thinning process in this manner, it is possible to effectively prevent the collapse of blanks without causing the disappearance of lines or the “haze” of various character images.

In addition, the number of times of the narrowing process is determined by, for example, the following equation according to the image size.

[Number of thinning processes] = [Vertical image size] / [Normalized size] By controlling the number of thinning processes in this way, it is possible to perform appropriate thinning processes on images of various sizes.

Specifically, for a character image in which the space between horizontal line segments is small and the normalized size is twice as large in the vertical direction, for example, a narrowing process from top to bottom and a narrowing process from bottom to top are performed. Apply alternately once. The reason why the thinning processing is alternately performed in the opposite direction is to disperse the influence of image distortion due to the thinning processing.

Since the thinning process described above is performed as preprocessing, the normalized character image generated by the OR processing unit 8 includes OR
The collapse of white space, which is a weakness of normalization, is unlikely to occur and OR
Since the normalization method is used, thin line segments disappear or “smear” unlike thinning normalization hardly occur. Therefore, it is possible to reduce erroneous recognition of characters due to crushing or “blurring”.

〔The invention's effect〕

As described above, according to the present invention, line segment thinning processing is performed as pre-processing of OR normalization processing, so that weaknesses of OR normalization are compensated for without losing their advantages, and blank spaces and line segments are eliminated. It is possible to obtain a normalized image without disappearance or “blurring”, and it is possible to collectively process various images simply by controlling the direction and the number of narrowing processes according to the nature and size of the target image.

[Brief description of the drawings]

FIG. 1 is a block diagram of a character recognition system according to the present invention,
FIG. 2 is an explanatory diagram of a thinning process in a direction from top to bottom, FIG. 3 is an explanatory diagram of a thinning process in a direction of bottom to top, FIG. 4 is an explanatory diagram of thinning-out normalization, and FIG. FIG. 6 is a diagram illustrating normalization, FIG. 6 is a diagram illustrating an example of thinning normalization, and FIG. 7 is a diagram illustrating an example of OR normalization. DESCRIPTION OF SYMBOLS 1 ... Original image input part, 2 ... Character extraction part, 3 ... Image memory, 4 ... Character image buffer, 5 ... Normalization processing part, 6 ... Narrowing processing direction / number control part, 7 ... Narrowing processing part, 8 ... OR Processing unit, 9: normalized image buffer, 10: character recognition unit.

──────────────────────────────────────────────────続 き Continued on the front page (58) Field surveyed (Int.Cl. ⁶ , DB name) G06K 9/42 G06K 9/44 JICST file (JOIS)

Claims (3)

(57) [Claims] In an image normalizing method for normalizing an image size, a direction of a thinning process is determined in accordance with a property of an image, a line segment thinning process is performed on an input image, and the line segment after the thinning process is processed. An image normalization method, which performs normalization on an image by a logical OR operation. 2. The image normalizing method according to claim 1, wherein
An image normalizing method, wherein a direction of a thinning process is determined according to the property of an image, and the number of times of the thinning process is determined according to the size of the image. 3. The image normalizing method according to claim 1, wherein the thinning process is performed alternately a predetermined number of times from the facing direction.