JPH07334621A - Character reading device - Google Patents

Abstract

PURPOSE:To provide a character reading device capable of appropriately eliminating undesired data such as the coarseness, noises of originals out of the binary image data and reading the characters with high accuracy. CONSTITUTION:This character reading device contains an image pickup device 2 which reads the images where the characters are described and a comparator 3 which binarizes the images read by the device 2 and outputs the binary image data. Furthermore, a 1st counter 4 to count the data detected out of all binary image data, a block dividing means 5 which divides the entire binary image data into plural blocks, a 2nd counter 6 which counts the detected data included in each of divided blocks, and a comparator means 7 which compares the count results of both counters 4 and 6 which each other and outputs the character data corresponding to the characters based on the result of comparison, are provided.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a character reader used for character recognition.

[0002]

2. Description of the Related Art Generally, as shown in FIG. 5, a character reading device of this type reads a character original 100 by an image pickup device 101, and output data of the image pickup device 101 is "1" or "0" by a comparator 102. Is converted into binary image data (binary image data). That is, the image data detected as the character area in the character original 100 is "1", and the image data not detected as the character area is "0".

However, in the binary image data obtained in this way, the image data is detected as a character area due to the roughness and noise of the character original 100, although it is not the original character area. Unnecessary data of "1" is included, and it is important to remove this unnecessary data from the viewpoint of improving the character reading accuracy.

Conventionally, as a method for removing unnecessary data of this kind, an isolated point removal method, a smoothing processing method, a labeling processing method and the like are known.

The isolated point removal method will be described. Binary image data including unnecessary data PD (FIG. 6A).
The isolated point removal process as shown in FIG. In other words, if there is "1" data in 4 or 8 directions, leave it as necessary data and leave "1".
If there is no data, it is replaced with "0" as unnecessary data. As a result, it is possible to obtain binary image data from which the unnecessary data PD has been removed, as shown in FIG.

The smoothing processing method is a method in which a low-pass filter is provided between the image pickup device 101 and the comparator 102 shown in FIG. 5, and unnecessary data, which is an isolated point, is removed as a high-frequency component by the action of this low-pass filter. .

Further, the labeling processing method is a method for detecting the continuity of binary image data, and for example, when the number of "1" data connections is three or less, it is removed as unnecessary data.

[0008]

However, the above-mentioned conventional unnecessary data removing method is effective when the amount of unnecessary data is relatively small, but when the amount of unnecessary data becomes extremely large, the unnecessary data disappears as isolated points. Therefore, there is a problem that it becomes difficult to completely remove unnecessary data.

The present invention has been made to solve the above-mentioned conventional problems, and its purpose is to accurately remove unnecessary data such as graininess and noise of an original from binary image data. It is to provide a character reading device. Another object is to provide a character reading device capable of highly accurate character reading.

[0010]

In order to achieve the above object, a feature of the present invention is that an image pickup device for reading an image in which characters are written and an image read by the image pickup device are binarized to be binary. In a character reading device having a comparator for outputting image data, a first counter for counting detection data in the entire binary image data;
Block division means for dividing the entire value image data into a plurality of blocks, a second counter for counting the detection data in each block divided by the block division means, and counts for the first and second counters The comparison means is provided for comparing the results and outputting the character data corresponding to the character based on the comparison result.

Further, in the above-mentioned invention, an average value calculating means for calculating an average value of the number of detected data per block in the plurality of blocks is provided based on the count result of the first counter, and the comparing means. It is preferable that the average value of the number of detected data is subtracted from the count result of the second counter, and only the block in which the subtraction result is larger than a predetermined threshold value is output as the character data.

Further, in the above invention, the predetermined threshold value may be fixed to zero.

Further, in the above-mentioned invention, the predetermined threshold value may be variable depending on the roughness of the image in which the character is written or the thickness of the character.

[0014]

The unnecessary data existing in the binary image data is considered to exist on average over the entire binary image data, and according to the present invention configured as described above, the first counter is
The detection data in the entire value image data is counted, and the second counter counts the detection data in each block divided by the block dividing means. Furthermore, the comparison means compares the count results of the first and second counters and outputs character data based on the comparison result. This makes it possible to easily remove unnecessary data from the binary image data even when the amount of unnecessary data is extremely large.

Further, the average value of the number of detected data per block is subtracted from the count result of the second counter, and only the block whose subtraction result is larger than a predetermined threshold value (for example, zero) is output as character data. By doing so, unnecessary data can be removed easily and accurately.

Further, by changing the predetermined threshold value according to the roughness of the image on which the characters are written or the thickness of the characters, the unnecessary data can be removed with higher accuracy.

[0017]

DETAILED 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 showing the configuration of a character reading apparatus embodying the present invention, and FIGS. 2 and 3 are diagrams showing an example of removing unnecessary data according to the present embodiment.

As shown in FIG. 1, this character reading device has an image pickup device 2 such as a CCD sensor for converting a character original 1 on which the character "8" is written into an electric signal,
The input side of the comparator 3 is connected to the output side of the image pickup device 2.

The electrical signal output from the image pickup device 2 is "0" or "1" with a constant threshold value by the comparator 3.
Is converted into binary image data D1. That is, the image data detected as the character area in the character original 1 is “1”.
It becomes data (detection data), and image data that is not detected as a character area becomes "0". At this time, the unnecessary characters due to the roughness and noise of the character original 1 are also binarized as "1" data together with the characters written on the character original 1. Here, the binary image data D1 is, for example, 25 in the horizontal direction.
It is output as a size of 6 pixels and 144 pixels in the vertical direction (see FIG. 2).

Further, the output side of the comparator 3 has 1
The block average value calculation unit 4 and the block division unit 5 are commonly connected. The one-block average value calculation unit 4 counts the number of detections of “1” data in the entire binary image data D1 (first counter), and counts this count value (eg, 286).
1) is divided by a preset block division number (for example, 192 blocks: described later) (2861 ÷ 192) to calculate the average number of detected “1” data (= 15) per block. However, the numbers after the decimal point are rounded off.

On the other hand, the block dividing unit 5 divides the entire binary image data D1 into blocks by the number of block divisions (192 blocks = horizontal 16 × vertical 12) (D2 in FIG. 2), and the block-divided binary. The image data D2 is sent to the block detection number counting unit 6. An enlarged view of one block at this time is shown at 11 in FIG. PD in the figure is unnecessary data.

Further, the block detection number counting section (second
Counter 6 counts the number of detected "1" data in each block. The number of detected "1" data in each block at this time is shown by 12 in FIG.

The output sides of the one-block average value calculating section 4 and the block dividing section 5 are connected to the input side of a difference detecting section 7 composed of an operational amplifier or the like. The difference detection unit 7 detects the average of “1” data per block calculated by the one-block average value calculation unit 4 from the detected number of “1” data of each block counted by the block detection number counting unit 6. The respective numbers (= 15) are subtracted, and the difference data D3 (see 13 in FIG. 3) is sent to the comparator 8.

This comparator 8 has a threshold value of "0" and detects only the block showing the plus data among the difference data D3 of each block as character data (see 14 in FIG. 3). Thereby, the binary image data (character data) OUT from which the unnecessary data PD is removed can be obtained.

The present invention is not limited to the above embodiment, and various modifications can be made. As a modification thereof, for example, the threshold value of the comparator 8 may not be fixed to "0" (that is, plus data), but may be changed according to the degree of roughness (roughness) of the document or the thickness of the character.

When the threshold value is changed according to the degree of graininess of the original document, the original document having a good roughness state (small unnecessary data) is ranked A (see FIG. 4A), and the intermediate one is ranked B (see FIG. 4B). )) And bad (more unnecessary data) are C ranks (see FIG. 4C), the threshold value of the comparator 8 is set as A rank> B rank> C rank. That is, the threshold value is lowered in consideration that the difference between the block of unnecessary data and the block of character data decreases as the roughness of the document deteriorates. As a result, the unnecessary data can be more surely removed, and the character can be read with high accuracy.

[0027]

As described above in detail, according to the present invention, a first counter for counting detection data in the entire binary image data and a block division for dividing the entire binary image data into a plurality of blocks. Means, a second counter for counting the detection data in each block divided by the block dividing means, the first and second
Since the comparison means for comparing the count results of the counter and outputting the character data corresponding to the character based on the comparison result is provided, even if the unnecessary data is remarkably large, it is unnecessary from the binary image data. The data can be completely removed, and the character reading accuracy is improved.

Further, in the above-mentioned invention, an average value calculating means for calculating an average value of the number of detected data per block in the plurality of blocks is provided based on the count result of the first counter, and the comparing means. Is configured to subtract the average value of the number of detected data from the count result of the second counter, and output only the block in which the subtraction result is larger than a predetermined threshold as the character data. With this configuration, unnecessary data can be removed accurately.

Further, in the above-mentioned invention, by fixing the predetermined threshold value to zero, it becomes possible to accurately remove unnecessary data with a simple structure.

Further, in the above-mentioned invention, by changing the predetermined threshold value according to the roughness of the image in which the characters are written or the thickness of the characters, unnecessary data can be removed with higher accuracy. It is possible to further improve the character reading accuracy.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a character reading device embodying the present invention.

FIG. 2 is a diagram illustrating an example of removing unnecessary data according to an embodiment.

FIG. 3 is a diagram illustrating an example of removing unnecessary data according to an embodiment.

FIG. 4 is a diagram showing a modification of the embodiment.

FIG. 5 is a diagram showing a configuration of a conventional character reading device.

FIG. 6 is a diagram for explaining an isolated point removal method.

[Explanation of symbols]

 1 character original 2 image pickup device 3 comparator 4 1 block average value calculation unit 5 block division unit 6 block detection number counting unit 7 difference detection unit 8 comparator D1 binary image data D2 block divided binary image data D3 difference data OUT character data

Claims (4)

[Claims] 1. An image pickup device for reading an image in which characters are written, and an image read by the image pickup device is binarized to be 2
In a character reading device having a comparator for outputting value image data, a first counter for counting detection data in the entire binary image data, and a block dividing means for dividing the entire binary image data into a plurality of blocks. Comparing the count results of the first and second counters with a second counter that counts the detection data in each block divided by the block dividing means, and corresponds to the character based on the comparison result. A character reading device provided with a comparing means for outputting the written character data. 2. An average value calculating means for calculating an average value of the number of detected data per one block in the plurality of blocks based on a count result of the first counter, and the comparing means includes the second value. 2. The average value of the number of detected data is subtracted from the count result of the counter and the only block in which the subtraction result is larger than a predetermined threshold is output as the character data. Character reader. 3. The character reading device according to claim 2, wherein the predetermined threshold value is fixed to zero. 4. The character reading device according to claim 2, wherein the predetermined threshold value is variable depending on the roughness of the image on which the character is written or the thickness of the character.