JP2985893B2 - Pattern recognition device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION Object of the Invention; (Industrial Application Field) The present invention extracts a feature amount of an input pattern based on density information of an input image when recognizing a paper sheet pattern, a character pattern, or the like. The present invention also relates to a pattern recognition device that recognizes a pattern based on the extracted input pattern feature amount.

(Prior Art) Conventionally, as a device for recognizing a sheet pattern, a sensor for reading the pattern is provided on a sheet conveyance path, and a pattern read by the sensor is compared with a reference pattern stored in advance. In some cases, the authenticity and type of paper sheets can be determined.

Further, in order to extract a pattern from a sheet, data read from the sensor is once binarized to separate the pattern from the background portion, and the binarized pattern and a reference pattern stored in advance are used. There is something to compare with.

(Problems to be Solved by the Invention) However, in the conventional pattern recognition method, fluctuations in the sensor system (a change in the light amount of the light source, a change in the sensitivity of the sensor, and a fluctuation in the gain of the amplifier) directly change the read data, and the fluctuations However, there is a case where recognition is erroneously performed because it directly affects the comparison with the reference pattern, and it is necessary to incorporate various correction methods and correction devices as a countermeasure. Further, the same problem occurs when the medium to be recognized (form) itself is faded as a whole or when the pattern on the medium is faint. Thus, there has been a strong demand for the emergence of a light and shade pattern recognition apparatus that uses a feature amount that does not require signal correction even for fluctuations in the sensor system and fluctuations in the recognition medium, and that is stable and improves recognition accuracy.

Further, when the read data is once binarized and the pattern is separated and extracted and compared with the reference pattern, a fluctuation occurs in the sensor system, the medium to be recognized is faded,
Further, when the pattern printed on the recognition medium is blurred, the pattern cannot be accurately separated and extracted because the binarization cannot be performed correctly. For example, the pattern portion is determined as a background portion that is not a pattern. In some cases, the patterns are separated, which causes a problem of misrecognition of the pattern.

The present invention has been made in view of the above-described problems, and an object of the present invention is to obtain shading information from an input image once when recognizing a paper sheet pattern or a character pattern, and to obtain the shading information. Provided is a pattern recognition device that extracts a feature amount of an input pattern by using the extracted feature amount of the input pattern and performs pattern recognition using the extracted feature amount of the input pattern, thereby enabling high-speed pattern recognition without lowering the recognition rate. It is in.

The present invention relates to a pattern recognition device, and an object of the present invention is to provide an image inputting means for inputting an image of a pattern to be recognized, and an image inputting means for inputting from the image inputting means. Pixel component extraction means for calculating a pixel component consisting of a density gradient direction component and a density gradient intensity component from each pixel of the image data by performing an arithmetic operation on each pixel of the image data using a plurality of masks composed of weighted pixels; A feature amount extracting unit that extracts, as a feature amount, a sum of density gradient intensities for each density gradient direction extracted by the pixel component extracting unit; and a storage unit that stores in advance a feature amount of a standard pattern to be a reference of a pattern to be recognized. Means, comparing and matching the feature amount of the pattern to be recognized extracted by the feature amount extracting means and the feature amount of the standard pattern stored in the storage means, On the basis of the comparison result is achieved by providing a comparison match means for determining whether the recognition pattern is any standard pattern.

(Operation) The present invention provides an apparatus for reliably recognizing a number, a bill pattern, or the like written on a check, a bill, or the like with a check writer.

According to the first aspect of the present invention, the image processing apparatus includes an image input unit, a pixel component extraction unit, a feature amount extraction unit, a storage unit, and a comparison and collation unit. The direction and intensity of the density gradient relating to shading are extracted from the image data, and the feature amount is extracted by the feature amount extracting means based on the direction, and the pattern to be recognized is recognized by comparing and comparing the feature amounts.

According to a second aspect of the present invention, the feature quantity extracting means of the first aspect is divided into an area dividing means and a feature quantity extracting means, and m × n
The density gradient information for each mesh is compared and compared as a feature amount.

(Example) An example of the present invention will be described below.

This embodiment is for recognizing the amount of money written on a check, a bill or the like, and the types of characters to be finally recognized are numbers "0 to 9" and yen symbols "@" and "*". There are 12 types of end digit codes such as. FIG. 2 (A) shows an example of a bill, and an amount entry column 1 for entering an amount with a check writer is provided in the center. FIG. 2B shows an example of a check, and an amount entry column 4 for entering an amount with a check writer is provided in the center. The present invention recognizes numbers, yen symbols, and last digit codes such as "*" to be entered by a check writer in the amount entry fields (1, 4) of these bills or checks.

First, one embodiment of the present invention will be described with reference to FIG.
An image input means 10 such as an image sensor inputs an image of a pattern to be recognized, pre-processes (smooths) image input information by a pre-processing means 11 and inputs the same to a pixel component extracting means 20. A pixel component composed of a density gradient direction component and a density gradient intensity component is extracted for each component. When the extraction of the pixel components is completed, the processing shifts to the processing by the feature amount extracting means. The feature amount extraction means 30 converts the image input by the image input means 10 into m
Xn meshes, and the sum of the density gradient intensity components for each of the density gradient direction components extracted by the pixel component extraction means 20 is extracted for each of the divided areas and used as the feature amount of the pattern to be recognized. The storage means 50 comprises a large classification dictionary 51 and a medium classification dictionary 52, and the comparison / matching means 40 compares the characteristic amounts extracted by the characteristic amount extraction means 30 with the characteristic amounts of the standard patterns stored in the storage means 50. Collation is performed, and a recognition result is output based on the comparison result.

Next, a recognition dictionary used for recognition according to the present invention will be described.

FIG. 3 is a schematic flow chart showing an operation of creating a recognition dictionary. First, in FIG. 3, an outline up to the extrusion of a standard feature for each standard pattern will be described. Step S1), smoothing is performed as preprocessing (step S2), and thereafter, a standard feature amount for each standard pattern is extracted (step S1).
3). Here, the steps of inputting the standard pattern and its image, smoothing processing, and extracting the standard feature amount will be described in detail.

The type of the standard pattern is as shown in FIG. 6 (A).
0 patterns have been created. Each of these 220 patterns is a standard pattern and contains most of the commonly used character forms. FIGS. 6 (B) to 6 (D) show some of the standard patterns in detail.
It is. For image input, the standard pattern of a predetermined size created in advance is read in 256 gray levels of 8 bits for each pixel in a rectangular range of 24 × 24 pixels, for example. FIG. 4A shows an example of the input image data, and the image data is subjected to a smoothing process as a pre-process. The smoothing process will be described below.

FIG. 7 is a flowchart showing an operation example of the smoothing process, and FIGS. 8 to 10 are explanatory diagrams thereof. In the smoothing processing, the input image data is smoothed by scanning the input image data with a 3 × 3 pixel smoothing mask shown in FIG. That is, first, as shown in FIG. 8, based on the density value of each pixel of the input image data, a histogram of the number of pixels of the density value is created (step S20), and the density value which becomes the peak is converted to the background density value. (BP) (step S21). this is,
This is because the normal character pattern has more background portions than character portions and the density value of the background portion is constant, so that the peak density value (BP) becomes the background density value. And the ninth
As shown in the figure, one pixel is arranged around a 24 × 24 pixel as a background density value to obtain image data of 26 × 26 pixels (step S2).
2). This is because a 3x3 pixel smoothing mask is 24x2
This is so that all the input pixel data in the four pixels can be operated. Next, the smoothing mask shown in FIG. 10 is scanned over the image data of 26 × 26 pixels (step S2).
3) The input image data is smoothed, and the result is used as subsequent image data. FIG. 4B shows an example of the image data at this stage. By performing the smoothing process in this way, the pattern formed by engraving the character with the check writer is filled with the density value of the character.
Character engraved patterns can be erased. Therefore, there is an advantage that a stable character feature can be obtained.

Next, a pixel component is extracted from the image data that has been subjected to the smoothing process as described above to extract a standard feature amount (step S3). The operation of extracting the pixel components will be described with reference to FIGS. 11 to 14.

FIG. 11 is a flowchart showing the extraction operation of the pixel components of the standard pattern. First, the density gradient intensity and the density gradient direction are obtained for each pixel using the Robinson operator on the image data subjected to the smoothing processing (step S30).
~ S32). A twelfth diagram indicate the Robinson operator, which consists in the eight the weight to each pixel of the 3 × 3 pixels are labeled mask M ₀ ~M _7, each mask M ₀ ~M ₇ thereof This is for extracting the gradient strength in the direction of the mask. For example, if scanning the image data by the mask M _0,
The gradient strength in the ↑ direction (EP ₀ ) is obtained for each pixel. Thus, it is possible to obtain by scanning the image data, density gradient strength for each direction for each pixel (EP ₀ ~EP ₇₎ using eight directions of the mask M ₀ ~M ₇ (step S3
0). The concentration gradient strength EP ₀ ~EP ₇ respectively mask M ₀
It corresponds to the ~M _7. The maximum value (EP _i ) is obtained from the density gradient intensities (EP _{0 to} EP ₇ ) thus obtained for each pixel (step S31). Then, the maximum value (EP _i ) is set as the density gradient intensity of the pixel, and the direction corresponding to the maximum value EP _i is set as the density gradient direction (step S32). Fig. 4 (C)
The image data thus obtained is shown in FIG.

Next, a binarization process is performed on the density gradient intensity obtained for each pixel thus obtained (steps S33, S40 to S40).
3). With reference to FIG. 13 (A), a method for determining a threshold value for binarization processing will be described. First, the maximum value of the density gradient intensity (EP _max ) is obtained from all the pixels (step S
33), it is checked that this maximum value (EP _max ) is within a predetermined range (step S40), and if there is any other value, an error is determined as an image input error (step S4).
1). Thereby, an image input error can be found at an early stage of the image processing. If the maximum value (EP _max ) is within the predetermined range, this 1/8 EP _max / 8 is set as a threshold (EP _TH ; hereinafter, referred to as “effective level threshold”) (step S4).
2). Then, the concentration gradient intensity "1" of the pixel is greater than the concentration gradient strength of each pixel (EP _i) is effective level threshold value (EP _TH), the pixel is smaller than the effective level threshold (EP _TH) Is set to "0" (step S4
3). Thus, the density gradient intensity of each pixel is binarized to “0” or “1”. This makes it possible to separate the pattern portion to be recognized from the background portion. FIG. 4D shows an example of the binarized image data.
FIG. 4E shows an example in which the direction of the concentration gradient is also displayed.

Here, the density gradient intensity is determined to be 0 or 1 by binarization to simplify, but the density gradient intensity itself for each pixel can be used as it is without binarization.
As shown in FIGS. 13B and 13C, the intensity of the concentration gradient can be quantized and used. In the case of FIG. 13 (B), the maximum value (EP _max ) is divided into eight levels, and the density gradient intensity is quantized into eight levels from 0 to 7. In the case of FIG. 13 (C), the cumulative addition of the histogram of the density gradient intensity is performed, and the maximum value thereof is divided into eight equal parts to quantize the density gradient intensity into eight stages of 0 to 7. If the density gradient intensity for each pixel is determined using the values quantized in multiple stages in this way, the accuracy is further improved.

Next, as shown in FIG. 14 (A), this binarized image data is applied to the analysis area to be recognized (mx ×
n) Divide equally and divide into mxn mesh areas (step S44). Here, 16 areas of 4 × 4 (SP _x [i]
[J] [k] i = 0 to 5, J = 0 to 3). Here, i is a horizontal section number, j is a vertical section number, and x is a standard pattern number. And
As shown in FIG. 14 (B), the sum of the density gradient intensities in each density gradient direction for each area (SP _x [i] [j] [k]; i = 0 to 3, j
= 0 to 3, k = 0 to 7) (step S45). However,
k is the direction of the concentration gradient. The value obtained in this way
SP _x [i] [j] [k] is a standard feature amount of the standard pattern, and is an i × j × k (= 128) -dimensional vector amount.
Through the above steps, the standard feature amount of the standard pattern is obtained. Then, for each standard feature, the feature is determined a predetermined number of times, for example, by using the same 10 standard patterns, and the average value is stored as a normal standard feature of the standard pattern (step S3).

In FIG. 3, the standard feature amount (SP _x [i] [j] [k]) is similarly obtained for all 220 standard patterns (step S4).

Next, the standard feature amount SP _x [i] [j] of each standard pattern
Create a large classification dictionary based on [k] (step S
Five). The contents of the large classification dictionary are as shown in Fig. 15, and the large classification dictionary uses the standard features of each standard pattern that have a certain similarity relationship as one small group (hereinafter referred to as cluster). , Is composed of multiple clusters. Specifically, a cluster is created based on the standard feature amount by a generally known cluster analysis method, for example, the Ward method. That is, the city block distance (D _ist ) indicating the similarity between the standard feature amounts of the standard patterns is obtained by the following equation (1) for all the combinations of the standard patterns.

Then, as a result of calculating the city block distances for all the combinations, those having the smallest distances are regarded as one cluster, and the average value of the standard feature amounts of the standard patterns belonging to the cluster is regarded as the feature amount of the cluster. Next, the city block distance of the feature amount is calculated again for all combinations of the cluster and the other standard patterns, and the one with the smallest distance is defined as one cluster.
Hereinafter, the same operation is repeated until the city block distance becomes a predetermined value or less, for example, 100 or less, or until the number of clusters becomes a predetermined number, for example, 50. In this way, the large classification dictionary is obtained by performing the clustering based on the similarity of the standard feature amount of the standard pattern by the Ward method.

Further, a feature amount is determined for each cluster. This is called a cluster feature. Cluster features (CSP _n [i]
[J] [k]; where n is a cluster number) is the average of the standard feature amounts of the standard patterns belonging to each cluster. As will be described later, during actual recognition, a large classification is performed based on the cluster feature, and the correspondence between the cluster feature and the cluster is a large classification dictionary. An example of the large classification dictionary at this stage is shown in FIG. On the other hand, the intermediate classification dictionary indicates the correspondence between each standard pattern and its standard feature amount. If you leave describes one Figure 15, for example in a cluster [14] Standard patterns belonging to this cluster "3" b3, a "5" b5, also in this cluster Cluster feature quantity CSP ₁₄ [I] [j] [k] is the standard feature SP _{b3 of b3} [i] [j] [k] and the standard feature SP _{b5 of b5}
It is the average value of [i] [j] [k].

Next, the operation of additionally registering a fluctuation pattern in the large classification basic dictionary in FIG. 15 will be described in step S10 and subsequent steps in FIG. The variation pattern is determined according to the reading accuracy of the recognition target medium (character pattern) of the recognition device in which the recognition dictionary is actually used. Specifically, assuming that the reading accuracy of the medium to be recognized by the recognition device is ± 4 degrees in the amount of rotation and ± 1 pixel (0.25 mm) in the amount of displacement, the variation pattern is, for example, −4 in the case of rotation. Degrees, -3 degrees, -2 degrees,
Ten kinds of variation patterns of -1 pixel, +1 degree, +2 degrees, +3 degrees, +4 degrees, and in the case of a shift, -1 pixel and +1 pixel are created for each of the standard patterns (220). An example is shown in FIG. 16 for the character "0". That is, FIG. 16 (A)
The variation pattern is shown as in FIG. Further, in this embodiment, the fluctuation pattern is created by separately performing the rotation and the shift.
It is also possible to add a variation pattern combining rotation and displacement, such as the displacement “−1 pixel”.

A pattern signal of a standard pattern corresponding to the variation pattern input as an image is input by the operation unit, and the variation pattern is input as an image (step S10). Smoothing processing is performed on the input image data in the same manner as when a standard pattern is input (step S11), and thereafter, a variation pattern feature amount (VP _x [i] [j] [k]) is obtained (step S12). The large classification dictionary is used to classify the large classification based on the feature amount. The large classification is a cluster standard feature (CSP _n [i] [j]) of each cluster in the large classification dictionary.
[K]) and the variation pattern features (VP _x [i] [j]
[K]) similarity between the (D _Istn) was calculated by the following equation (2) (step S13), and the highest degree of similarity (Dist _n is minimum) is determined as belonging to the cluster (step S1
Four).

Then, it is checked whether or not the standard pattern corresponding to the fluctuation pattern has already been registered in the determined cluster (step S15). If not, the pattern is additionally registered in the cluster (step S15). , S17). The registration is performed only with the pattern number corresponding to the standard pattern. If the pattern is already registered, the above operation is performed for the next variation pattern. The above-described operation for additionally registering a variation pattern is performed for all variation patterns corresponding to the standard patterns (220 patterns), and the process ends when all of the variation patterns are completed. In addition, when the variation pattern is additionally registered, only the pattern number of the standard pattern is additionally registered in the cluster, and the cluster feature amount used for the large classification is not updated. This eliminates the need for large classification in all standard patterns again when a new reference pattern different from the existing 220 is added, prevents the cluster features from slowing down, and resembles the cluster's wollast features. This is to increase the gender difference.

Next, at the time of additionally registering the variation pattern, the weight for each area for each pattern used in the detailed classification to be described later is calculated (steps S18, S18AS19). This will be described. This is because even if a change occurs during image input, it is necessary to check in advance which areas are less affected by the change. This calculation operation will be described with reference to FIG.

Standard pattern number x processed when creating the large classification dictionary
N variation pattern features (VP _{x, p} [i]
[J] [k]) (P: Character variation pattern numbers 1 to n) Average feature amount for each 4 × 4 mesh area (i, j)
▼ [i] [j] [k] and its feature σ _x ^{2 for} each area
[I] [j] is calculated by the following equations (3) and (4),
The inverse of the feature variance σ _x ² [i] [j] for each area is weighted by WT
It is stored as _x [i] [j]. When the calculation of the feature amounts for all the variation patterns is completed, the average feature amount ▼▼ and the feature amount variance σ of the calculated variation pattern feature amounts VP _x [i] [j] [k] for each mesh area (i, j). the _x ² is calculated according to the following (3) and (4) (step S5
0). Then, the reciprocal of the feature amount variance σ _x ² is determined by the weight WT _x [i].
[J] (step S51), and is registered in the weight dictionary (step S52). The area where the weight WT _x [i] [j] is large is a stable area that has little influence even if a change occurs at the time of image input.

FIG. 18 is an example of a large classification dictionary at this stage in which a variation pattern is additionally registered.
For example, in cluster No. 14, four standard pattern numbers “e3, e5, b8, o8” are additionally registered. In addition,
The large classification dictionary shown in FIG. 18 is used for actual pattern recognition. Here, the large classification dictionary indicates the correspondence between clusters, which are similar character groups, and the cluster features of the clusters. The large classification based on the large classification dictionary refers to the feature amount of the character to be recognized. This is a process of comparing with a cluster feature amount to determine to which cluster the character to be recognized belongs.

Next, the character pattern recognition operation using the above recognition dictionary will be described with reference to the block diagram of FIG.

First, for example, an image of the check amount entry column 4 shown in FIG. 2B is input by the image sensor of the image input means 10.
The amount entry column 4 is a predetermined position, and the image sensor obtains the shade density in this column in 256 gradations.
FIG. 5 (A) shows an original image of the money amount entry column 4. FIG. The image data thus obtained is smoothed by the preprocessing means 11 to perform smoothing. As a result, as shown in FIG. 5B, the character extracting means 17 extracts characters based on the image data subjected to the smoothing processing. To cut out characters, the image data is scanned by the above-mentioned Robinson operator, and a left component in the horizontal direction (FIG. 4 (F))
Then, cut-out coordinates are extracted from a histogram of pixels having a rightward component (FIG. 4 (G)). As a result, a leftward histogram of FIG. 5 (E) and a rightward histogram of FIG. 5 (F) are obtained. In other words, the place where the extreme change occurs in the lower part of the right histogram is the right end of the character,
In the leftward histogram, a location that changes extremely at a low position is the left end of the character. Similarly, the vertical direction is cut out in the vertical direction by using a histogram of an image having an upward component (FIG. 4 (H)) and a downward component (FIG. 4 (I)). In this embodiment, since it is one line in the horizontal direction, it is only necessary to perform the operation for the entire money amount entry column 4. If each of the horizontal cutouts is cut out in the vertical direction, it is possible even when they are not aligned in the horizontal direction. Then, the central point of the region cut out in this way is obtained, and a region of 24 × 24 pixels centering on this point is set as a character cut-out region. This is because the feature amount of the above-described standard pattern is obtained as 24 × 24 pixels, so that it is matched.

FIG. 5 (C) is an image of the density gradient, and FIG. 5 (D) shows an image obtained by binarizing the density gradient. FIG. 5 (E) is a leftward histogram, and a left end line 100 which defines the left end of each character is obtained. FIG. 5 (F) is a rightward histogram, which defines the right end of each character. Right end line 20
0 is obtained, and characters are cut out by the left end line 100 and the right end line 200 as shown in FIG.

Next, a feature value (hereinafter, referred to as an input pattern feature value) is extracted by the feature value extracting means 32 for each area divided as described above. Input pattern features (IP
_m [i] [j] [k]; where m is a cutout number, i is a horizontal section number, j is a vertical section number, and k is a density gradient direction number). In the same manner as described above, the density gradient intensity and the density gradient direction are obtained and binarized for each image using the Robinson operator's mask on the image data of the extracted character, and the character extraction area is further reduced to 4 ×. 4 are divided into 16 areas (IP _m [i] [j]), and the total number of pixels in each density gradient direction is calculated for each area. The 128-dimensional vector amount thus obtained is the input pattern feature amount. Then, when the extraction of the input pattern features for all of the extracted characters is completed, the large classification unit 41 uses the large classification dictionary 51 based on the input pattern features.
The pattern recognition proceeds in three stages, namely, large classification in the above, middle classification in the middle classification section 42, and detailed classification in the detailed classification section.

That is, first, a large classification is performed in order to identify a cluster. The large classification is similarity between each cluster standard feature quantity CSP _n [i] [j] [k] for each cluster in the large classification dictionary in FIG. (D _istn ) is calculated by the following equation (5), the most similar one (the one with the smallest D _ist ) is obtained, and it is determined that the character belongs to the cluster.

Note that, as described above, the standard pattern in the cluster of the large classification dictionary fails in the large classification because the additional registration is made in the cluster by the feature amount when the fluctuation occurs due to the rotation shift at the time of image input. Never. When the cluster to which the cut-out character belongs is determined by the large classification, the intermediate classification is performed by the intermediate classification unit 42 next. Middle classification, the input pattern feature quantity _{IP m [i] [j]} [k] and the standard feature value SP _x of all standard patterns belonging to the cluster [i] [j]
The similarity (D _istx ) with [k] is calculated by the above equation (5), and two candidates which are similar to each other by a predetermined value or more and which are the most similar (from the smaller D _istx ) are extracted as candidates. .

At this time, when there is no similar thing (when _Distx does not have a value equal to or less than a predetermined value), it is determined that the character is not in the standard pattern, and a numeral “0 to 9” or a yen sign “￥”, “*” Assume that the character is not the last digit sign. If there is only one similar pattern, it is determined to be the standard pattern. Normally, two candidates are extracted. In this case, it is determined whether or not the two standard patterns have the same character. For example, when "0" of the standard pattern "a0" and "0" of "f0" of the same character type are extracted, the most similar standard pattern is set, and the character is set to "0".
Is determined. This means that the characters are finally "0-9",
This is because it is only necessary to know which of “￥” and “*”, and the standard pattern may be either “a0” or “f0”. When two standard patterns of different characters are extracted, the process proceeds to the detailed classification in the detailed classification unit.

The detailed classification compares the extracted standard patterns of the two candidates, extracts a stable dissimilar area from the feature amounts of each pattern of each pattern, and performs pattern matching only on the area. A specific description will be given using the explanatory diagram of FIG. 20 according to the flowchart of FIG.

As shown in the FIG. 20 (A) and (B), and in middle classification extracted as two candidate reference pattern x ₁ and x _2. Then, the difference D between the extracted standard patterns
_iff [i] [j] is calculated for each area (4 × 4) by the following (6)
It is determined by the formula.

The expression (6) means the expression obtained by multiplying the degree of difference between the feature amounts of two candidate character standard patterns in a certain area by the stability (weight) of the feature amount in that area. A measure for extracting a stable area with high dissimilarity between two candidate characters even when a change occurs is expressed. Then, has been required to WT _x [i] [j] when additional registration of the variation patterns in the dictionary creation process in (6). FIG. 20 (C) shows a specific example in which the degree of difference for each area is obtained by equation (6), and three areas having a large degree of difference are extracted from the degree of difference for each area thus obtained. I do. In the example of FIG. 20 (C), _Diff [1] [1], _Diff [0] [1],
_Diff [3] [0] is extracted, and the hatched area in FIG. These three areas are areas where the standard patterns are very different from each other, and are areas where the influence caused by the fluctuation occurring during image input is small. Therefore, the input pattern is similar to either standard pattern only for this area. If you look at it, you can more accurately determine. First,
Only for the three extracted areas, the similarity AD _istx between the feature quantity of the input pattern and the feature quantity of the two candidate standard patterns
[I] [j] is obtained by the following equation (7).

The sum of the differences between the three areas is calculated as the area similarity.
AD _istx [i] [j]. In the example of FIG. 20, each area similarity can be expressed by the following equations (8) and (9).

The two standard patterns thus obtained (x ₁ and
x ₂ ) and the similarity of the area similarities AD _istx1 and AD _istx2 are compared, and the more similar one (the smaller AD _ist ) is determined as the character of the input pattern. This character recognition operation for the input pattern is performed for all of the extracted characters, and then the recognition operation ends.

At this stage, the amount information entered in the amount entry column of the check or the like can be recognized.

In the above embodiment, the Ward method of the cluster analysis is used at the time of creating the large classification time. However, the present invention is not limited to this, and any method can be used as long as it can be classified by a certain similarity between patterns. Further, in the above embodiment, the average of the feature amounts of the patterns in the cluster is used as the cluster feature amount. However, the average may not be used, and one feature amount of the pattern in the cluster may be used as the cluster feature amount. In short, any feature amount that can represent a pattern in a cluster can be used. Furthermore, in the above-described embodiment, the variation pattern is created in advance and the variation pattern feature is extracted by inputting the image. However, the present invention is not limited to this method. Image data of the variation pattern may be created by performing image processing such as rotation and displacement on the image data at that time, and the variation pattern feature amount may be extracted.

In the above embodiment, the character pattern is recognized by extracting characters from a form such as a check. However, in FIG. 1, step S12 is omitted, and the feature amount of a pattern such as a bill is stored in a dictionary. Then, it can be used as it is for pattern recognition of paper sheets such as banknotes. Further, the feature quantity extracting means of the above embodiment divides the input image into m × n areas and obtains the sum of the density gradient intensities for each density gradient direction component for each area. , The sum of the density gradient intensities for each of the density gradient direction components may be used as the feature amount.

As described above, according to the present invention, the density gradient direction and the intensity relating to the density are once obtained from the input image data, and the characteristic amount is extracted and recognized based on the density gradient information. It is hardly affected by fluctuations of the system (changes in the light intensity of the light source, changes in the sensitivity of the sensor, fluctuations in the gain of the amplifier), and is almost unaffected even when the recognition medium is faded or the pattern is blurred. Even in this case, the pattern can be reliably recognized. Further, since the input image is divided into m × n areas and the feature amount of the pattern is obtained based on the density gradient information for each area, it is possible to perform comparison and collation for each local part of the pattern, thereby achieving higher accuracy. Pattern recognition becomes possible.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention.
FIG. 2A is a diagram showing an example of a bill, FIG. 2B is a diagram showing an example of a check, FIG. 3 is a flowchart showing a method for creating a recognition dictionary according to the present invention, and FIGS. (I) and FIG. 5 each show a specific image example, and FIG. 6 (A)
To (D) show examples of standard patterns, FIG. 7 is a flowchart showing an example of smoothing processing, FIGS. 8 to 10 are diagrams for explaining smoothing, and FIG. 12 to 14 are diagrams for explaining the extraction operation, FIG. 15 is a diagram showing an example of a large classification dictionary, FIG. 16 is a diagram for explaining a standard pattern, FIG. FIG. 18 is a diagram showing an example of the middle classification dictionary, FIG. 18 is a diagram showing an example of the middle classification dictionary, FIG. 19 is a flowchart showing an operation example of the detailed classification, and FIG. 20 is a diagram for explaining the detailed classification. is there. 1,4 ... Amount entry column, 2,5 ... Information column, 3,6 ... Amount information column, 10 ... Image input unit, 11 ... Preprocessing unit, 20 ... Pixel component extraction unit, 30 ... ... Feature extraction means, 31 area division means, 40 comparison and comparison means, 50 storage means.

──────────────────────────────────────────────────続 き Continued on the front page (58) Fields investigated (Int. Cl. ⁶ , DB name) G06K 9/00-9/82 G06T 7/00 G07D 7/00

Claims (4)

(57) [Claims] An image input means for inputting an image of a pattern to be recognized, and a plurality of masks each comprising a weighted pixel for each pixel of the image data input from the image input means, thereby calculating the image data. A pixel component extracting means for extracting a pixel component composed of a density gradient direction component and a density gradient intensity component from each pixel, and a total sum of density gradient intensities for each density gradient direction extracted by the pixel component extraction means is extracted as a feature amount. Characteristic amount extracting means, storage means in which the characteristic amount of a standard pattern serving as a reference for a recognition slip pattern is stored in advance, characteristic amount of the pattern to be recognized extracted by the characteristic amount extracting means and stored in the storage means. A comparison / matching unit that compares and matches the feature amounts of the standard patterns, and determines which standard pattern is the recognized pattern based on the comparison result. Pattern recognition apparatus characterized by comprising. 2. The method according to claim 1, wherein the characteristic amount extracting means divides the input image into a plurality of m × n areas, and for each of the divided areas, each of the density gradient direction components extracted by the pixel component extracting means. 2. The pattern recognition apparatus according to claim 1, wherein a sum of density gradient intensity components is extracted and used as a feature amount of the ratio recognition pattern. 3. The eye pattern recognition apparatus according to claim 1, wherein the recognition pattern is a paper sheet pattern. 4. The pattern recognition device according to claim 1, wherein the recognition pattern is a character pattern.