JPH0137790B2 - - Google Patents

Description

[Detailed description of the invention] [Technical field of invention] The present invention relates to a pattern recognition method.

[Background technology of the invention and its problems]

There are a pattern matching method, a stroke analysis method, a feature analysis method, a composite similarity method, etc. to recognize printed characters, but the pattern matching method is generally widely adopted.

Character recognition using this pattern matching method involves capturing an image of the character along with its background using a camera, and then converting the image information obtained from this camera to a binarization circuit.
In this method, the quantized information is converted into binary quantized information, and this quantized information is successively compared with a plurality of dictionary patterns stored in a dictionary, and the dictionary pattern that most closely matches is used as the recognition result. However, this method
The disadvantage is that it is sensitive to changes in the line width of characters and often causes recognition errors. In other words, the line width of printed characters generally changes depending on printing conditions, so for example, if the line width is thick like character 1 shown in Figure 1A, this is stored in the dictionary shown in Figure 1B. When superimposed on the dictionary pattern 2, a large mismatched area 3 is generated as shown in Fig. 1C.
When superimposed with other dictionary patterns, a clear distinction cannot be obtained, leading to recognition errors.

In order to prevent such recognition errors, it is possible to store a large number of dictionary patterns with different line widths for the same character in the dictionary, but increasing the number of dictionary patterns in this way requires a large memory capacity. Not only that, but also the time required for recognition becomes longer.

Another method is to compare the quantization information obtained by converting the image information with the dictionary pattern.
There is also a method of weighting characteristic parts of characters for determination, but this method requires a more complex and larger device than the above-mentioned method, and its manufacturing cost is higher. Furthermore, there is a pattern recognition method disclosed in Kyoritsu Shuppan Co., Ltd.'s Information Science Structure E19.2 "Character/Graphic Recognition Machine" pages 44 to 49, but this method uses a pre-stored reference mask. There is a problem in which it is mistakenly recognized that the line width of the input image is not equal to the line width of the input image.

[Purpose of the invention]

An object of the present invention is to provide a pattern recognition method that can stably and quickly recognize a pattern even if the line width of a pattern such as a character or figure changes.

[Summary of the invention]

Regarding the binarized pattern information, after obtaining a first polarity signal and a second polarity signal which is an inversion of this first polarity signal, the first polarity signal is the same as that stored in advance. The signal of the second polarity is matched with the memory output of the same second polarity stored in advance, and the signal of the second polarity is matched with the memory output of the first polarity and the second polarity. This method obtains a pattern recognition method that determines an input pattern from the results of .

[Embodiments of the invention]

Hereinafter, an embodiment in which the present invention is applied to character recognition will be described based on the drawings.

FIG. 2 is a circuit configuration diagram of a character recognition device.
Reference numeral 10 denotes a camera that captures an image of the characters printed on the substrate surface together with the substrate surface that is the background of the characters, for example.
The ITV camera 11 receives the signal of a predetermined character area from among the ITV camera output signals, A/D converts the image information sent from the camera 10 corresponding to the captured image, and converts it into binary data. A binarization circuit 12 converts the input quantization information into quantization information of a first polarity, for example, a positive image, sent from the binarization circuit 11.
A first dictionary 13 has a plurality of dictionary patterns (positive masks) of polarity stored in a memory in advance; This is a first comparator that sequentially compares, for example, the background portions of characters of a plurality of dictionary patterns, and calculates the degree of matching between the two for each dictionary pattern. 15 is an inversion circuit for inverting the polarity of the quantized information sent from the binarization circuit 11 to a second polarity; 16 is for identifying the inverted quantized information (negative) sent out from the inversion circuit 15; a second dictionary in which a plurality of reversed (negative) dictionary patterns are stored in advance in memory;
Reference numeral 17 denotes a character portion signal (negative image) of quantized information of second polarity obtained by inverting the polarity of the input signal sent from the inverting circuit 15, and a plurality of dictionaries of second polarity stored in the second dictionary 16. For example, by sequentially comparing the character part of the pattern (negative mask) and matching,
This is a second comparator that calculates the degree of matching between the two for each dictionary pattern. Reference numeral 18 is for determining input characters, and the logical product of the degree of coincidence calculated by the first comparator 13 and the degree of coincidence calculated by the second comparator 17 is calculated for each identical character in each dictionary pattern. This is a judgment circuit that calculates the

Assuming that a signal consisting of a plurality of characters with a lower reflectance than the base surface is printed on the substrate surface, the character information signal is taken in from the camera 10 by sampling or other means, and the transmitted image information is also As shown in FIG. 3A, the character portion 21a is composed of a character pattern 22 having a lower reflectance than the background portion 20a. Based on the image information sent from the camera 10, the binarization circuit 11
As shown in the figure, the character pattern 22 is converted into a binary pattern 23 (positive image) in which the character portion 21b is "0" and the background portion 20b is "1". This is used to obtain first polarity pattern information. This binary pattern 23 is converted into information of the second polarity by an inverting circuit 15, that is, as shown in FIG.
It is converted into an inverted binary pattern 24, that is, a negative image, in which c is set to "1" and the background portion 20c is set to "0".

The first dictionary 12 stores in advance a plurality of dictionary patterns (positive masks) of the first polarity for comparison with the binarization pattern 23 sent out from the binarization circuit 11, as shown in FIG. 3D. In each dictionary pattern 25, the character part 21d is "0" and the background part 20d is "1", as in the binarization pattern 23.
and the line width of the characters is the above-mentioned binary pattern 23.
It consists of a pattern that is thicker than the line width of the character, for example, the maximum thickness that can be recognized as a character. 1st
The comparator 13 compares the characters in the image captured by the camera 10, that is, the camera 1 corresponding to this image.
This is a circuit for checking the characters included in the image information sent from 0, and is a circuit that checks the characters contained in the image information sent out from 0, and combines the binarization pattern 23 sent out from the binarization circuit 11 and the plurality of characters stored in the first dictionary 12. Dictionary pattern 2
Matching, that is, the degree of coincidence is calculated for the background parts 20b and 20d with 5. That is, as shown in FIG. 3F, the binarized pattern 23 is sequentially superimposed on a plurality of dictionary patterns 25, and the number of pixels in which the background portions 20b and 20d are both "1" is determined for each dictionary pattern 25. In this case, the degree of agreement is
The number of pixels in the background part 20b of the binarized pattern 25 is
When n ₁ and the number of pixels in the background portion 20d of the dictionary pattern 25 are N ₁ , it is expressed as n ₁ /N ₁ , and when both patterns 23 and 25 completely match, n ₁ /N ₁
= 1, and the value decreases as the number of unmatched portions increases.

The second dictionary 16 stores a plurality of dictionary patterns to be compared with the inverted binary pattern 24 sent from the inverting circuit 15, and as shown in FIG. 3E, each dictionary pattern 27 is Similarly to the binarization pattern 24, the character part 21e is "1",
The background portion 21e is set to "0", and the line width of the characters is thinner than the line width of the characters of the inverted binary pattern 24, for example, the minimum thickness that constitutes a clear character. The second comparator 17 is a circuit that checks the skeleton of a character included in the image information sent from the camera 10 in response to the calculation of the degree of coincidence in the first comparator 13, and The degree of coincidence between the character parts 21c and 21e between the inverted binary pattern 24 sent from the second dictionary 16 and the plurality of dictionary patterns 27 stored in the second dictionary 16 is calculated. That is, as shown in FIG. 3G, the inverted binarized pattern 24 is sequentially superimposed on a plurality of dictionary patterns 27, and the number of pixels in which the character portions 21c and 21e are both "1" is determined for each dictionary pattern 27. . The degree of matching is expressed as n ₂ /N ₂ , where n ₂ is the number of pixels in the character portion 21c of the inverted binary pattern 24, and N ₂ is the number of pixels in the character portion 21e of the dictionary pattern 27. If patterns 24 and 27 match perfectly, n ₂ /
N ₂ = 1, character part 2 of dictionary pattern 27
The value increases as the portion that does not match 1e increases.

The degrees of coincidence calculated by the first and second comparators 13 and 17 are sent to a sequential determination circuit 18. In other words, dictionary patterns 25 and 27 regarding the same character
The degrees of coincidence obtained by comparing the two are sequentially transmitted. The determination circuit 18 sequentially accepts the degree of coincidence and sequentially calculates their logical product n ₁ /N ₁ ×n ₂ /N ₂ . And dictionary pattern 2 that yields the maximum logical product value
It is assumed that the characters are taken with 5 or 27.

For symbols consisting of multiple characters, the entire symbol can be read by repeating the above operation for each character.

A binarized pattern 23 corresponding to the character pattern 22 of the image information sent from the camera 10 as described above, and a binarized pattern 23 that is binarized in the same way as this binarized pattern 23 and whose line width is the binarized pattern. pattern 2
Multiple dictionary patterns 25 that are thicker than the line width of the characters in 3
The background parts 20b and 20d are compared sequentially.
In addition to determining the degree of matching, a binary pattern 24 which is inverted with respect to the binary pattern 23, which is binarized in the same manner as this inverted binary pattern 24, and whose line width of the character is inverted as described above. The character portions 21c, 2 are sequentially compared with a plurality of dictionary patterns 27 that are thinner than the line width of the characters of the binarized pattern 24.
1e is determined, and these background parts 20b, 2
If a character is judged by calculating the logical product of the degree of coincidence of 0d and the degree of coincidence of character parts 21c and 21e, one character will be checked from different viewpoints, and even if the line width of the character changes, there will be no misidentification. It is possible to perform stable recognition without fear of

Furthermore, even if one character is checked twice, the degree of correspondence between the background parts 20b and 20d of the character and the degree of correspondence between the character parts 21c and 21e are processed by binarization, and are determined simultaneously in real time. Since it can be sent to the circuit 18, character recognition can be processed at high speed.

Furthermore, since the recognition process is performed on a single sheet of paper, the apparatus can be made compact.

An embodiment of the present invention has been described above, but when capturing an image of a subject using a camera, a symbol consisting of a plurality of characters may be simultaneously imaged, and each character may be extracted and processed one by one from the image information obtained as a result. However, the characters constituting the symbol may be imaged one by one with a camera and processed.

Further, the degree of coincidence sent from the first and second comparators may be calculated by logical product in real time as shown in the above embodiment, but it may also be calculated after being stored in a memory. Furthermore, in the above embodiment, a case has been described in which recognition is performed by first performing background verification and then character verification; however, it is also possible to perform recognition by first performing character verification and then performing background verification. good.

Furthermore, in the above embodiment, an example was explained in which the present invention was applied to character recognition, but it goes without saying that the present invention can be applied to any signal that can be treated as pattern information, such as graphic information or audio information.

〔Effect of the invention〕

According to this invention, pattern information of the first polarity and pre-stored memory contents of the first polarity are matched, and pattern information of the second polarity and pre-stored memory contents of the second polarity are matched. Since the pattern is recognized by matching the pattern information, stable recognition is possible even if the line width of the pattern information changes.

[Brief explanation of drawings]

Figures 1A to 1C are explanatory diagrams of a conventional character recognition method using a pattern matching method, Figure 2 is a circuit diagram of a character recognition device of the present invention, and Figures 3A to G are illustrations of a character recognition method of this invention. FIG. 10...Camera, 11...Binarization circuit, 12...
...first dictionary, 13...first comparator, 15...inverting circuit, 16...second dictionary, 17...second comparator,
18... Judgment circuit, 22... Character pattern, 23
... Binarization pattern, 24 ... Inverted binary pattern, 25, 27 ... Dictionary pattern.

Claims (1)

[Claims] 1. A means for binarizing input pattern information, a first polarity signal for the binarized pattern information, and a second polarity signal obtained by inverting the first polarity signal.
means for obtaining a signal of the same polarity as the signal of the first polarity; means for matching the signal of the first polarity with a memory output of the same first polarity stored in advance; and means for matching the signal of the second polarity with a memory output of the same first polarity stored in advance. A pattern recognition method comprising: means for performing matching with a memory output of a second polarity; and means for determining an input pattern from the results of matching according to each of the first and second polarities. 2. The pattern recognition method according to claim 1, wherein the matching based on the first polarity is matching regarding a background portion, and the matching based on the second polarity is matching related to pattern information.