JPH05282494A - Binary coding device for image data - Google Patents

Abstract

PURPOSE:To provide the binary coding device for image data which obtains a binary- coding result having mean line width closest to an optimum line width set value for character recognition by performing a binary coding process only once. CONSTITUTION:The binary coding device is provided with a line width calculation part 18 finding mean line widths for plural threshold values at the same time when multilevel image data are coded into binary data on the basis of the threshold values, a threshold value determination part 13 which selects one line width closest to the optimum line width set value among the line width calculation results based upon the threshold values and outputs the binary coding threshold value corresponding to the line width, a multilevel image delay buffer 19 wherein multilevel image data within the line width detection range are temporarily stored until the threshold value is determined, and a binary coding part 14 which codes the multilevel image data in the multilevel image delay buffer into binary data sequentially by using the threshold value. Consequently, the binary coding threshold value which provides the mean line width closest to the optimum line width set value for character recognition is determined by single-time scanning and then the multilevel image data in the multilevel image delay buffer 19 are coded into binary data.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a character recognition device or the like in which an average line width after binarizing image data obtained from multi-valued image data input means is an optimum line width setting value for character recognition. The present invention relates to a binarization apparatus that determines a binarization threshold value so that the binarization threshold is closest to, and particularly to an image data binarization apparatus that dynamically binarizes.

[0002]

2. Description of the Related Art A conventional binarization method using an average line width is
It is performed by an apparatus as shown in FIG. In FIG. 2, the multi-valued image data obtained from the multi-valued image data input means 21 is stored in the multi-valued image scanning buffer 22.
Then, during the first binarization process, multi-valued image data in the line width detection setting range is sequentially taken out from the multi-valued image scanning buffer 22, and based on the threshold initial value output from the threshold determination unit 23, It is binarized by the binarization unit 24. The binarization result is stored in the binary image memory 25, and at the same time, sequentially extracted in the scanning order in the form of 3 × 3 pixels by the 3 × 3 pixel extraction unit 27, and the line width using the 3 × 3 pixel data. The calculator 28 obtains the average line width information. On the other hand, in the threshold value determining unit 23, the average line width information thus obtained is compared with the optimum line width setting value for character recognition, and the current binarization threshold value is reduced in the direction in which the difference between the two values becomes smaller. Is increased or decreased and output. The second and subsequent binarization processes are performed using this binarization threshold value, and the binarized image data is obtained in the binary image memory 25 by the threshold value at which the average line width is closest to the line width setting value. Repeat the binarization process up to. Examples of this kind of method include, for example, "Character reading control method; Yamada, Tada, Iida, The Institute of Electrical Communication Research, Proceedings No. 36, February 1981, Nippon Telegraph and Telephone Public Corporation.
Institute of Electrical Communication, pp. 55-56 ”.

[0003]

In the above prior art, in order to obtain the optimum binarization result, the average line width information is compared with the line width setting value, and the maximum and threshold values are set until the difference between them is minimized. There is a problem that it is necessary to repeatedly increase and decrease the binarization threshold value a number of times, and it takes processing time for the repetition.
Further, the processing time required for binarization is not constant, and it is difficult to perform efficient dynamic binarization processing using a pipeline. An object of the present invention is to improve such a problem and to obtain a binarization result having an average line width closest to an optimum line width setting value for character recognition in one binarization process. Can
Another object of the present invention is to provide an image data binarization device capable of reducing the amount of memory.

[0004]

In order to achieve the above object, an image data binarizing device of the present invention binarizes multivalued image data inputting means and multivalued image data with a plurality of threshold values. A means for simultaneously obtaining the average line width for each threshold value (a black point counting section, a contour point counting section, a black dot / contour point number table, a line width calculating section, and a line width calculating section), and a plurality of From the line width calculation result based on the threshold of, the line width closest to the optimum line width setting value is selected, and a threshold value determination unit that outputs a binarization threshold value corresponding to the line width and the threshold value The memory (multi-valued image delay buffer) that temporarily stores the multi-valued image data for the line width detection setting range and the threshold value are used to sequentially convert the multi-valued image data in the multi-valued image delay buffer into binary values until it is determined. And a binarization unit for converting the binarization threshold value. It is characterized by performing binarization after.

[0005]

In the present invention, the line width calculation unit and the threshold value determination unit determine the binarization threshold value which is the average line width closest to the optimum line width setting value for character recognition by one image scanning. After that, the binarization unit binarizes the multi-valued image data for the line width detection setting range temporarily stored in the multi-valued image delay buffer. As a result, the binary image data with the optimum binarization threshold value can be obtained by performing the binarization process once, so that the processing time can be shortened. Further, since the processing time required for binarization is constant, efficient dynamic binarization processing can be performed using a pipeline. Furthermore, since the binarization process is completed by one scan, a multi-value image scanning buffer for holding the original image data is unnecessary, and the amount of memory can be reduced accordingly.

[0006]

DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram of a binarization device according to an embodiment of the present invention. In FIG. 1, 11 is multi-valued image data input means for obtaining image data to be processed as multi-valued data, and 17 is 3 × from multi-valued image data output from the multi-valued image data input means 11. A 3 × 3 pixel extraction unit that sequentially extracts multi-valued image data in the scanning order in the form of 3 pixels, 1
Reference numeral 8 denotes a line width calculation unit that simultaneously calculates respective average line widths for a plurality of thresholds by using the 3 × 3 pixel multi-valued image data output from the 3 × 3 pixel extraction unit 17, and 13 a line width calculation. A threshold determination unit that selects one closest to the set line width from the plurality of average line width information obtained from the unit 18 and outputs a binarization threshold corresponding to the line width, and 19 is a threshold determination unit. The multi-valued image delay buffer for storing multi-valued image data for the line width detection setting range until the binarization threshold is determined by 13, and 14 is the binarization obtained by the threshold value determination unit 13. A binarization unit that sequentially binarizes the multivalued image data stored in the multivalued image delay buffer 19 using a threshold value,
Reference numeral 15 is a binary image memory for storing the image data binarized by the binarization unit 14, and 16 is a binary image memory 15
It is a character recognition unit that performs character recognition by inputting the binarization result in the above.

Next, the line width calculator 18 will be described in detail. FIG. 3 is a configuration diagram of the line width calculation unit in the embodiment of the present invention. The line width calculation unit 18 of FIG. 1 determines whether the central multi-valued image data G _{0 of} 3 × 3 elements is a black point for a plurality of threshold values and counts the black point counting unit 181 and G _0. A contour point counting unit 182 that determines whether a point is a point by using the 8-neighbor multi-valued image data G _{1 to} G ₈ of G ₀ , and a black point / contour that simultaneously stores the number of black points and the number of contour points for a plurality of thresholds The score table 183 and black dots for each threshold
The line width calculation unit 184 calculates the average line width from the number of contour points, and the line width information table 185 that stores the average line width calculation result for each threshold. Each line width information in this line width information table The data is the output of the line width calculator. Here, the average line width W is calculated by W = 2 × B / B _E, where B is the number of black dots and B _E is the number of contour points. The contour point determination method in the contour point counting unit 182 is described in detail in Japanese Patent Publication No. 3-7180, for example. With such a configuration, the average line width when binarizing the multi-valued image data with a plurality of binarizing thresholds can be obtained at the same time, so that the binarizing unit 14 performs the binarizing process once. After obtaining the binary image data in which the average line width is closest to the line width setting value, the character recognition unit 1 in the latter stage
Can be sent to 6.

[0008]

According to the present invention, it is necessary to perform the conventional iterative processing when determining the binarization threshold value such that the average line width is closest to the optimum line width setting value for character recognition. Since there is no error, the binarization result is always obtained after the delay of the time required for constant line width detection setting range scanning, so the binarization processing time becomes constant, and efficient dynamic binarization processing using a pipeline is performed. Can be done. Further, since it is not necessary to perform scanning a plurality of times as in the conventional case, a multivalued image data holding buffer (multivalued image scanning buffer) is unnecessary, which is useful for reducing the memory amount.

[0009]

[Brief description of drawings]

FIG. 1 is a configuration diagram of a binarizing device according to an embodiment of the present invention.

FIG. 2 is a diagram showing a configuration example of a conventional binarization device.

FIG. 3 is a configuration diagram of a line width calculation unit according to an embodiment of the present invention.

[Explanation of symbols]

 11 multi-valued image data input means 13 threshold value determination section 14 binarization section 15 binary image memory 16 character recognition section 17 3 × 3 pixel extraction section 18 line width calculation section 19 multi-valued image delay buffer 21 multi-valued image data input section 22 Multi-value image scanning buffer 23 Threshold value determination unit 24 Binarization unit 25 Binary image memory 26 Character recognition unit 27 3x3 pixel extraction unit 28 Line width calculation unit 181 Black dot counting unit 182 Contour point counting unit 183 Black dot / contour Point table 184 Line width calculator 185 Line width information table

Claims (1)

[Claims] 1. An image data binarization apparatus comprising: means for inputting multivalued image data; and a memory for storing a binarization result of the input multivalued image data. Means for simultaneously obtaining the average line width for the threshold value when binarizing with the threshold value and the line width calculation result by the plurality of threshold values, one line width closest to the predetermined line width setting value is selected and selected. Means for outputting a binarization threshold value corresponding to the line width, a memory for temporarily storing multi-valued image data for a line width detection setting range until the threshold value is determined, and the threshold value,
A binarizing device for image data, comprising means for sequentially binarizing multi-valued image data in the memory.