JPH0743754B2 - Character image binarization device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial application] The present invention relates to a sensor (image input sensor) for scanning a character image region to be read in an optical character reading device.
For the image signal obtained from, the character that is used to binarize, such as "0" for the signal representing the other part of the recording medium on which the character is recorded and "1" for the signal representing the character pattern part. The present invention relates to an image binarization device.

[Conventional technology]

FIG. 4 is a circuit diagram showing a conventional example of a character image binarization device. In the figure, 1 is an image input sensor, 2 is a comparator,
Reference numeral 3 is a resistor, 4 is a capacitor, 5 is a variable constant voltage source, and 6 is a resistor.

FIG. 4A is a waveform diagram of signals at various parts in FIG.
FIG. 4B is an explanatory diagram showing an example of an image to be read. In FIG. 4B, 7 is a recording card, 8 is a bar code, and S is the scanning direction.

The circuit operation for binarizing the input image signal will be described with reference to FIGS. 4, 4A, and 4B.

That is, in FIG. 4, the image input sensor 1 is
Raw image input signal (a) obtained when the bar code 8 on the recording card 7 as shown in the figure is scanned along the S direction
Is input to the (+) input side of the comparator 101 and, on the other hand, the raw signal (a) is passed through a low-pass filter composed of a resistor 3 and a capacitor 4 to remove harmonic components and A reference signal (b) used as a reference is created and is input to the (-) input side of the comparator 101. (-)
A constant voltage is previously applied to the input side from the variable constant voltage source 5 via the resistor 6.

In this way, the line image signal (a) obtained by scanning the bar code 8 at the (−) input side of the comparator 2 is
As described above, a smooth signal from which harmonic components have been removed is obtained. Therefore, by comparing the levels of the signals (a) and (b) in the comparator 2, the binarized signal (c) is converted into the barcode 8
Is obtained as the read output of.

By the way, in the conventional binarization apparatus as described above,
An image consisting of multiple parallel line segments, such as a barcode,
This is effective when scanning with the image input sensor from the direction perpendicular to it, but when the scanning direction of the image input sensor is parallel to the line segment forming the bar code, the length of that line segment Depending on the size, it may or may not be binary.

As an extreme example, in the case of a straight line extending in a direction coinciding with the scanning direction of the image input sensor, this cannot be binarized. When the object to be read is not a line segment such as a bar code but a character, the line segment forming the character includes a component parallel to the sensor scanning direction, a component perpendicular to the sensor scanning direction,
Oblique components, point components, and the like are mixed, and if the conventional binarizing device as shown in FIG. 4 is used to binarize the same, its operation becomes unstable.

[Problems to be solved by the invention]

Therefore, in the present invention, the object to be read is not an image composed of line segments aligned in parallel to one direction like a bar code, but the line segments constituting it are different from the scanning direction of the sensor like ordinary characters. Targeting an image pattern that consists of line segments that form various angles, a stable binarization operation is performed on the image signal obtained from the sensor in the pattern portion and other portions, and unevenness in position such as illumination, ground, and sensor sensitivity is detected. Regardless of the above, making possible is the problem to be solved. Therefore, it is an object of the present invention to provide a character image binarization apparatus that enables the above.

[Means and Actions for Solving Problems]

To achieve the above object, the present invention utilizes a two-dimensional local memory. The two-dimensional local memory is, for example, a two-dimensional image signal obtained by picking up an image with a two-dimensional image pickup device such as a television camera, pixelized and quantized, and two-dimensionally stored in the memory. The pixel data in an arbitrary desired local area of the memory can be taken out (note that as a two-dimensional local memory, the present invention is a convenient one in which the possible local area can be freely changed. −223697).

Therefore, in the present invention, as the local area in the two-dimensional local memory, a local area having a width larger than the width of the line forming the target character pattern is considered, and the central pixel and a plurality of peripheral pixels in the area can be read as the representative pixels. In this way, the gray value of each of these representative pixels is examined, the maximum value and the minimum value are obtained, and the intermediate value between the maximum value and the minimum value is calculated, and the gray value of the central pixel and this intermediate value are calculated. According to whether or not the difference (absolute value) is greater than or equal to a predetermined value, "0" or "1" of the central pixel is determined, and the binarization is performed.

That is, in the local area assumed in the present invention, when the central pixel is a pixel on a line segment forming a character pattern, the peripheral pixels always include the pixel of the portion corresponding to the background, that is, the ground. thinking.

〔Example〕

 Next, an embodiment of the present invention will be described with reference to the drawings.

FIG. 5 is an explanatory diagram showing an example of a character image that the binarizing apparatus according to the present invention considers to be binarized and a waveform of an image signal obtained by scanning the image.

That is, the image signal to be binarized by the binarizing device according to the present invention is an image signal obtained by scanning general characters (alphabet, hiragana, etc.) as shown in FIG. .

For example, the waveform of the image signal obtained by optically scanning the line x ₁ along the main scanning direction is as Sx, and is obtained by optically scanning the line y ₁ along the sub scanning direction. The waveform of the image signal is like Sy, but unlike the image signal obtained by scanning the code bar as shown in Fig. 4A, the DC level on the line segment that constitutes the character is the character position. It varies greatly due to, and depending on the conventional binarization device,
It will be appreciated that it is difficult to binarize as described above.

FIG. 1 is a block diagram showing the overall construction of an optical character reader (OCR) incorporating a binarizing device according to the present invention.

In the figure, reference numeral 11 denotes a sensor (image input sensor) that optically scans a character image written on a recording medium such as paper and outputs an image signal, specifically, an image reading unit of a television camera or a fax machine. It may be considered to correspond to. 12
Is an A / D converter that quantizes the image signal from the sensor 11 and outputs it as a digital signal of a plurality of bits (for example, 6 bits, 64 gradations), 13 is the above-mentioned two-dimensional local memory, and 14 is 2
The digitization unit, 15 is a character recognition unit.

The two-dimensional local memory 13 is, as described above, the Japanese Patent Application No. 60
-It is described in detail in the specification of No. 223697,
Here, to the extent necessary for understanding the present invention,
This will be described with reference to FIG.

FIG. 2 is an explanatory diagram showing an example of a local area that can be taken in the two-dimensional local memory 13 and a pixel arrangement representing the area.

That is, a partial character image area of (M dots) × (N dots) is designated as a local area, and a central pixel D in the area is designated.
Consider ₀ (represented by a black circle) and surrounding freely selectable peripheral pixels D ₁ (represented by a white circle) as representative pixels that represent the local area, and simultaneously quantize the respective pixels. The two-dimensional local memory can be read and referenced. Therefore, in the scanning screen, if the central pixel is shifted by one dot along the scanning direction, the local area centered on the central pixel also moves, so that every pixel is centered on the entire screen. The surrounding pixels when considered as a pixel are considered, and the quantized value of each pixel at that time can be known.

FIG. 3 is a block diagram showing details of the binarizing unit 14 (one embodiment of the present invention) in FIG. In the figure, 14
1 is a maximum value calculation circuit, 142 is a minimum value calculation circuit, 143 is an average value calculation circuit, 144 is an absolute difference calculation circuit, and 145 is a comparison circuit.

Returning to FIG. 1, the overall circuit operation of the optical character reader will be briefly described. An image signal obtained by scanning the character image by the image input sensor 11 is quantized by the A / D converter 12 for each pixel and input to and stored in the two-dimensional local memory 13. Next, the quantized values (shading values of pixels on the recording medium) of the central pixel as a representative pixel representing the local area and its peripheral pixels are read from the two-dimensional local memory 13, and the binarizing unit 14 reads the quantized values. , The central pixel is binarized, and the result is sent to the character recognition unit 15. The character recognition unit 15 performs character recognition on the target image scanned by the sensor 11 from the binarized dot (pixel) pattern obtained as described above.

Next, the circuit operation of the binarization unit 14 as an embodiment of the present invention will be described with reference to FIG.

The binarization unit 14 has the structure shown in FIG. 3 as already described. The maximum value calculation circuit 141 simultaneously refers to the output of the two-dimensional local memory 13, that is, the center pixel of interest and the grayscale values of a plurality of peripheral pixels located at a predetermined distance from the pixel, and determines the maximum grayscale value among them. know. The minimum value calculation circuit 142 similarly knows the minimum gray value. When the target image is characters written in black on a white background such as paper, the maximum value calculation circuit 141 sets the maximum value of the gray value of the color of the paper in the vicinity of the center pixel of interest and the minimum value calculation circuit 142. Is set and calculated as the minimum gray value of the character color (black).

The average value calculation circuit 143 calculates an intermediate value between the maximum value and the minimum value calculated as described above, that is, an average value. The absolute difference calculation circuit 144 calculates the absolute difference between the average value from the average value calculation circuit 143 and the grayscale value of the center pixel of interest. By using the absolute difference in this way, it is possible to perform binarization in which the character portion and the background portion can be distinguished regardless of whether the black character is on the white background or the white character is on the black character.

Next, the comparison circuit 145 does not recognize a slight change in the gray value in the background portion (ground portion) with respect to the output from the absolute difference calculation circuit 144 in consideration of the S / N and recognize it as a character portion. Then, a predetermined threshold value α is subtracted, and binarization is performed depending on the presence or absence of the remaining value.

When the absolute difference is calculated by the absolute difference calculation circuit 144,
A background portion near a character in which a character is included in a local area is binarized in the same manner as the character when the binarization is performed with the background portion as a central pixel. However, this is merely a recognition of a thick character, and when the local area is entirely in the background portion, it becomes "0" and can be discriminated from the character portion, so there is no practical problem.

As described above, the stable binarization that allows the character portion to be effectively distinguished from the ground portion can be performed without being confused by the gradual change in gray value that may exist in the ground portion.

By the way, the expression of the binarization method according to the present invention is as follows.

The center pixel of interest is p ₀ , and the peripheral pixels are p ₁ , p _2, ... P ₈ . p ₀ ~
p ₈ is a pixel representing the local area AL.

If the maximum gray value in the local area AL is p _max , then p _max = max {p ₀ , p ₁ , p ₂ ... p ₈ } If the minimum gray value in the local area AL is p _min , then p _mix = min {p ₀ , p ₁ , p ₂ ... p ₈ } Let p _m be the average gray value of p _max and p _min , and _let the gray level difference between p _m = (p _max + p _min ) / 2 p _m and the center pixel of interest p ₀ be Let p _d be p _d = │p ₀ −p _m │, where p _b is the binarized output at the center pixel of interest p ₀ , and α is the set input (threshold value), p _b is as follows.

{0 ... ( _Pd <α} _pb = {1 ... ( _pd ≧ α} [Effect of the Invention] According to the present invention, a width and a width that are sufficiently larger than the width of the line segment that constitutes a character can be obtained. The central part in the local area and the background part in the local area while referring to the gray values of a plurality of peripheral pixels,
The binarization is performed by knowing the lightness level of the character part and the degree of change of the lightness level of the central pixel with respect to the average value, and thus the fluctuation of the lightness level of the background part in the image and the lightness of the character part. There is an advantage that stable binarization can be performed so that the character part and the background can be distinguished from each other even if the level changes.

Further, according to the present invention, since the absolute difference is used, it is possible to perform stable binarization without changing the threshold value for binarization even when white characters and black characters are mixed. it can.

[Brief description of drawings]

FIG. 1 is a block diagram showing the overall structure of an optical character reading device (OCR) incorporating a binarizing device according to the present invention.
FIG. 4 is an explanatory diagram showing an example of a local area that can be taken in the two-dimensional local memory and an example of pixel arrangement representing the area. FIG. 3 is a block diagram showing the details of the two-dimensional local memory 13 in FIG. 1, and FIG. A circuit diagram showing a conventional example of a character image binarization device, FIG. 4A is a waveform diagram showing waveforms of its input signal and output signal (binarized output), and FIG. 4B is an example of an image to be read. FIG. 5 is an explanatory diagram showing an example of a character image that the binarizing apparatus according to the present invention considers to be binarized, and an explanatory diagram showing a waveform of an image signal obtained by scanning the image. is there. Explanation of symbols 1 ... Image input sensor, 2 ... Comparator, 3 ... Resistor, 4 ... Capacitor, 5 ... Variable constant voltage source, 6 ... Resistor, 7 ... Recording card, 8 ... Bar code, 11 ... Image input sensor, 12 ... A / D converter, 13 ... Two-dimensional local memory, 14 ... Binarization section, 15 ... Character recognition section, 141 ... Maximum value calculation circuit, 142 ... Minimum value calculation circuit,
143 ... Average value calculation circuit, 144 ... Absolute difference calculation circuit, 145 ... Comparison circuit,

Claims (1)

[Claims] 1. A target character image area is decomposed into pixels,
A character image input sensor that scans in the main scanning direction and sub-scanning direction to detect and output a character image signal for each pixel, and an A / A that outputs the character image signal output from the sensor by quantizing A D (analog / digital) converter and a quantized signal from the converter are input, and a width larger than the width of a line segment forming the character image with the pixel of interest as the center pixel in the character image area. A two-dimensional local memory that sequentially outputs, for each pixel of interest, quantized signals for the central pixel and a plurality of peripheral pixels in the two-dimensional local area set to a wide area, the local area Calculation for obtaining the average value by selecting the maximum value and the minimum value from the quantized signals for the central pixel and the peripheral pixels for each, and for obtaining the absolute difference between the average value and the quantized signal of the central pixel. Means and the absolute difference And a certain set value, and a binarizing means for making a binarization judgment of the central pixel according to the magnitude relationship between the two, and a character image binarizing apparatus.