JP3193135B2 - Dot pattern dot position detection method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for detecting a dot position of a dot pattern based on image information.

[0002]

2. Description of the Related Art An example of a conventional dot pattern identification method will be described with reference to the drawings.

[0003] The identification of the dot pattern is roughly performed using four means shown in FIG.

The first image input means, as shown in FIG. 2, divides an image of a dot pattern input from a camera into pixels in the x and y directions by an A / D converter. The pixels are converted into digital images having discrete density values and stored in an image memory. In the following means,
By processing the image data stored in the memory, the dot pattern is identified.

As a second means for cutting out each pattern, as shown in FIG. 3, the projection value (total value of density values) is obtained by a projection process of summing pixel densities on a straight line in the x direction. If the length of a portion where is continuously equal to or more than a threshold value is substantially equal to the length of the pattern in the y direction, it is determined that the pattern exists and the upper and lower ends of the pattern are determined. Similarly, when the length of the portion where the projection value in the y direction is continuously equal to or greater than a certain threshold is substantially equal to the length of the pattern in the x direction, it is assumed that the pattern exists and the left and right ends of each pattern are determined. To determine. In the example of FIG. 3, the projection threshold for the y-direction clipping is thy, and the upper end position of the pattern is y1.
, The lower end position is indicated by y2. The projection threshold for the cutout in the x direction is thx, the left end position of each pattern is x11,
x12, x13 and the right end position are indicated by xr1, xr2, xr3. The circumscribed rectangle is obtained from the upper, lower, left and right ends of each pattern as shown in FIG.

The third dot detecting means will be described. The circumscribed rectangle obtained by the cutting means of each pattern is mesh-divided such that each dot is included in one mesh as shown in FIG. Next, the presence or absence of dots in each mesh is determined. The method is as follows. In each mesh, if a pixel having a density threshold or more is equal to or more than a certain determination number, a dot is included in the mesh. Shall exist. For example, when the density value of each pixel in one mesh is as shown in FIG. 6, if the density threshold value is 50 and the number of determinations is 8, there are dots in this mesh, If the threshold value is 70 and the number of determinations is 8, there are no dots in this mesh.

Assuming that a mesh where dots exist is 1 and a mesh which does not exist is 0, 0/1 data in a compressed format as shown in FIG. 7 is obtained for the image shown in FIG. .

The fourth pattern identification means has the above-mentioned 0/1 reference data for each identification pattern in advance, and determines which 0/1 data obtained from each segmented pattern corresponds to which identification data. The pattern is identified by referring to whether it matches well with the reference data of the pattern.

[0009]

However, in the above-mentioned conventional technique, as shown in FIG. 8, when the dot diameter becomes large or when the brightness of the image changes,
The dot detection in the mesh is affected by the density threshold, the number of pixels for determining the presence of a dot in the small area, or the position of the circumscribed rectangle of the character. As shown, correct dot detection may not be performed. In view of the above, an object of the present invention is to enable detection of a dot which is considered to affect the above-described dot detection and is not easily affected by various fluctuation amounts.

[0010]

In order to achieve the above object, the first invention of the present application detects a point where the density of a pixel in an image becomes a peak in comparison with surrounding pixels, and this peak is detected. It is characterized in that the presence or absence of a dot in each small area is determined based on whether or not a point exists in each small area, and the position of the dot is detected.

In order to achieve the above object, the second invention of the present application reduces the dot diameter by performing a process of replacing a pixel in an image with an operation result of the pixel and surrounding pixels, thereby reducing a dot diameter. It is characterized in that the presence or absence of a dot in each small area is determined based on the attribute of each data of the reduced diameter data for each small area to detect the position of the dot.

According to a third aspect of the present invention, in order to achieve the above object, a dot detection area smaller than the small area is provided in each small area, and the presence or absence of a dot is determined only by the dot detection area. It is characterized in that the presence or absence of a dot in each small area is determined based on the detected position of the dot.

[0013]

According to the first aspect of the present invention, utilizing the phenomenon that the dot density peaks in the vicinity of the center of the dot, each pixel at the point where the pixel density peaks in comparison with the surrounding pixels. The presence or absence of a dot in each small area is determined based on the attribute for the area, so the dot diameter and position, the brightness of the image, the density threshold for separating the dot part from its background, the small area It is possible to perform dot detection that is not easily affected by a change in the number of pixels for determining the presence of a dot, the position of an area including a pattern, or the like.

According to the second aspect of the present invention, utilizing the phenomenon that the dot density peaks near the center of the dot,
Since the presence / absence of a dot in each small area is determined based on the attribute for each small area of the data with the reduced dot diameter, the dot diameter and position, the brightness of the image, and the dot portion are separated from the background. In this case, it is possible to perform dot detection that is not easily affected by the density threshold value, the number of pixels for determining the presence of a dot in a small area, and the position of the area including the pattern.

According to the third aspect of the present invention, a dot detection area smaller than the small area is provided in each small area, and the presence or absence of a dot is determined only by the dot detection area. Since the presence / absence of a dot is determined, if the peak point of the dot density and the dot detection area are made to match, the dot diameter and position, the brightness of the image, the dot portion can be determined from the background. It is possible to perform dot detection that is not easily affected by a density threshold value at the time of separation, the number of pixels for determining that a dot exists in a small area, a change in the position of an area including a pattern, and the like.

[0016]

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 10 shows a dot pattern to be read. Characters are represented by dots in five rows and five columns. The patterns to be discriminated are 10 kinds of characters from Arabic numerals 0 to 9. Further, here, the density of the dot portion is assumed to be brighter than that of the background portion, but the same concept can be applied to the case where the density of the dot portion is darker than the background portion.

Hereinafter, a procedure for reading characters will be described.

(1) An image of a character input from a television camera is divided into pixels to obtain a digital image. At this time, it is assumed that the density of each pixel is quantized to an integer value from 0 to 255 so that a bright portion has a large value. The obtained image is subjected to a shading correction by a filtering process as a pre-process, thereby removing illumination unevenness and the like.

(2) Projection values are calculated for the preprocessed image in the vertical and horizontal directions, and a portion where the projection value is continuously large is regarded as a character portion, and a rectangular cutout circumscribing the character is cut out. Do. FIG. 11 shows an example in which a rectangle is cut out.

(3) The cut-out rectangle is divided into 5 vertical and 5 small rectangles corresponding to each dot of the dot pattern. FIG. 12 shows an example in which the rectangle shown in FIG. 11 is divided into small rectangles. By detecting a dot in each small rectangle, and setting a small rectangle in which a dot exists as 1 and a small rectangle in which no dot exists as 0, pattern data in a compressed form as shown in FIG. 13 is obtained.

(4) For each of the identification patterns, data of 1/0 of this format is stored in advance, and characters are identified by referring to which identification pattern data the obtained pattern matches well. . In the reference method, an exclusive OR is performed between values corresponding to each mesh position between 0/1 data obtained in advance from a pattern to be identified and reference data, and the sum of the obtained values is calculated. When it is closest to 0, it is assumed that the pattern to be identified is a pattern corresponding to the reference data at that time. For example, if the reference data for each of the identification patterns “2”, “3”, and “4” is as shown in FIG. 14 and the data obtained from the pattern in the image is as shown in FIG. And the data of the exclusive OR between the pattern “1” and the identification patterns “2”, “3”, and “4” are as shown in FIG. 16, and the sum of the values for each pattern is 4, 2, 12, Become. Therefore, the data in FIG. 15 is identified as “3”.

The present invention relates to the dot detecting method (3). FIG. 17 illustrates a dot detection method, which implements one of three methods (corresponding to the first invention), (corresponding to the second invention), and (corresponding to the third invention). Hereinafter, three embodiments of dot detection will be described separately.

In the first embodiment, when the pixel density of the dot portion has a peak near the center as shown in FIG. 18, a point where the density is peak in a rectangle circumscribing the character is detected. . As a detection method, it is assumed that the magnitude relationship between eight pixels adjacent to each other is examined, and a pixel which is not smaller than any of those adjacent pixels is a peak point. FIG. 19 shows an image of a dot character by a density value. The solid line in the figure is a small rectangle divided by the above method. FIG.
Peaks detected by this method for the image shown in FIG. 9 are indicated by circles in FIG.

Next, the concentration values at these peak points are tested. That is, a peak point whose density value is smaller than a predetermined density threshold value is excluded. Threshold 80
In FIG. 21, the remaining peak points are indicated by circles.

Finally, a small rectangle having a dot and a small rectangle not having a dot are determined from the attributes of the remaining peak points for the small rectangle. In the example shown in FIG. 19, FIG. 22 shows data of a pattern in a compressed form obtained by setting a small rectangle having dots to 1 and a small rectangle not to exist to 0.

In the second embodiment, when the density of the dot portion has a peak near the center as shown in FIG. 18, the density of the peripheral portion is reduced while keeping the density of the peak portion. Thereby reducing the dot diameter. Hereinafter, the method will be described. First, a maximum value filter process is performed to replace the pixel density of the processing target with the maximum value of the density value of the total m × n pixels of vertical m pixels × horizontal n pixels centering on the processing target pixel, and Is stored in another image memory (where m and n are integers). FIG. 23 shows an image after the maximum value filter processing when m = n = 3 with respect to the image before the processing shown in FIG. By performing an inter-image calculation process of Expression (1) using the obtained image,
The density of the peripheral portion can be reduced while maintaining the density of the peak portion. FIG. 24 shows an image after the image-to-image calculation of Expression (1).

G (x, y) = f (x, y) × 2−fma (x, y) (f (x, y) × 2> fmax (x, y) = 0 (f (x, y) × 2 ≦ fmax (x, y) (1) where g (x, y) is the image after processing, f (x, y) is the image before processing, and fmax (x, y) is the maximum value. It is an image after filter processing.
Also, x and y indicate plane coordinates in the image.

In the finally obtained image, if the number of pixels included in a small rectangle of a pixel whose density value is larger than a predetermined density threshold value is greater than a predetermined judgment number, a small value is set. Assuming that a dot exists in the rectangle, a small rectangle in which the dot exists and a small rectangle in which the dot does not exist are determined. FIG.
In the image shown in (1), when the density threshold value is 80 and the number of judgments is 1, the compressed rectangular pattern is obtained by setting the small rectangle having dots to 1 and the small rectangle not to exist to 0. 1/0 data is shown in FIG.

The second embodiment is different from the first embodiment in that
In addition to being realized by a combination of commonly used filtering modules, there is an advantage that it is hardly affected by bright dot-like noise and the like.

In the third embodiment, in a small rectangle,
An area for detecting a dot indicated by oblique lines in FIG. 26 is provided. That is, it is determined whether or not there is a dot included in the small rectangle only in this detection area. The determination of the presence / absence of a dot is based on the presence of a dot in a small rectangle if the number of pixels having a density value greater than a predetermined density threshold within a detection area is equal to or greater than a certain determination number. Is determined, a small rectangle in which a dot exists and a small rectangle in which a dot does not exist. For example, for the image shown in FIG.
Only one pixel at the center of 3 × 3 pixels included in the small rectangle is set. On the other hand, the density threshold is 80 and the number of judgments is 1
FIG. 27 shows data of 1/0 of a pattern in a compressed format obtained by setting a small rectangle having a dot to 1 and a small rectangle not having a dot to 0.

[0032]

According to the present invention, the diameter and position of a dot, the density threshold for separating a dot portion from its background, the number of pixels for judging the presence of a dot in a small region, and an area including a pattern , It is possible to detect a dot that is not easily affected by a change in the position of the dot, and to detect the dot position accurately.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of a dot pattern identification device.

FIG. 2 is a configuration diagram showing an image input unit.

FIG. 3 is a diagram showing cutout means for each pattern to be identified.

FIG. 4 is an explanatory diagram of a state where a dot pattern is cut out by a circumscribed rectangle.

FIG. 5 is an explanatory diagram of a state in which a dot pattern cut out from a circumscribed rectangle is further divided into small rectangles.

FIG. 6 is a diagram illustrating an example of a pixel density in one small rectangle.

FIG. 7 is a diagram illustrating an example of compressed format data obtained as a result of dot detection.

FIG. 8 is a diagram showing an example in which the dot diameter is uneven.

FIG. 9 is a diagram showing data in a compressed format obtained as a result of detecting the dots of FIG. 8 in a conventional example.

FIG. 10 is an explanatory diagram of dot characters to be identified.

FIG. 11 is an explanatory diagram of a state where a dot pattern is cut out by a circumscribed rectangle.

12 is an explanatory diagram of a state where the dot pattern cut out by the circumscribed rectangle in FIG. 11 is further divided into small rectangles.

FIG. 13 is a diagram showing 1/0 data in a compressed format obtained as a result of dot detection in the small rectangle of FIG. 12;

FIG. 14 is an explanatory diagram of reference data used for character identification.

FIG. 15 is a diagram showing 1/0 data obtained as a result of performing dot detection from image data.

FIG. 16 is a diagram showing an exclusive OR between the elements shown in FIGS. 14 and 15;

FIG. 17 is a schematic explanatory diagram of a dot position detecting method of the present invention.

FIG. 18 is an explanatory diagram of an image having a peak near a dot center.

FIG. 19 is an explanatory diagram of the density of pixels of a character to be read in the first embodiment of the present invention.

20 is an explanatory diagram of a detection state of a density peak point of the image of FIG. 19;

21 is an explanatory diagram of points remaining as a result of further performing threshold processing on the density peak points detected in FIG. 20;

22 is a diagram showing 1/0 data obtained from a dot position detected from a mesh attribute of a point remaining in FIG. 21.

FIG. 23 is an explanatory diagram of the image density after the maximum value filter processing in the second embodiment of the present invention.

FIG. 24 is an explanatory diagram of an image density in the second embodiment.

25 is a diagram showing a dot obtained from the image of FIG. 24 by dot detection.
It is a figure which shows the data of / 0.

FIG. 26 is an explanatory diagram of a dot detection area according to the third embodiment of the present invention.

FIG. 27 is a diagram showing 1/0 data obtained by dot detection in the dot detection area of FIG. 26;

Claims (3)

(57) [Claims] 1. An image processing method according to claim 1, wherein an area including a dot pattern is determined based on image information of a pattern composed of dots.
Divide into multiple small areas determined by the dots of the pattern ,
Next, in the dot position detection method of the dot pattern that detects the position of the dot from the attribute of the dot for each small area, the point where the density of the pixel in the image becomes a peak in comparison with the surrounding pixels is detected. Each peak point is
A dot position detection method for a dot pattern, wherein the presence or absence of a dot in each small area is determined based on whether or not the dot exists in the small area to detect the position of the dot. 2. An area including a dot pattern is defined as a reference dot based on image information of a pattern composed of dots.
Divide into multiple small areas determined by the dots of the pattern ,
Then over attributes of dot for each small area, the dot position detection method of the dot pattern for detecting the position of the dot, is subjected to the pixels in the image, processing for replacing with the calculation result of its pixel and surrounding pixels By
A dot of a dot pattern, characterized in that the dot diameter is reduced, the presence or absence of a dot in each small area is determined based on the attribute of the reduced data for each small area, and the position of the dot is detected. Position detection method. 3. An area including a dot pattern is defined as a reference dot based on image information of a pattern composed of dots.
Divide into multiple small areas determined by the dots of the pattern ,
Next, in a dot position detection method of a dot pattern for detecting a dot position from a dot attribute for each small area, a dot detection area smaller than this small area is provided in each small area, and a dot exists only in the dot detection area. A dot position detecting method for a dot pattern, comprising: judging the presence or absence of a dot and detecting the presence or absence of a dot in each small area based on the judgment to detect the position of the dot.