JPS6228889A - Information recognizing device - Google Patents

Abstract

PURPOSE:To input characters, codes and graphics such as alphabetical small letters and codes, which may be varied in each writer by forming a discrimination part for discriminating a category having similarity more than a prescribed value as a proposed recognized result. CONSTITUTION:When a discriminating circuit 3 obtains the feature of input pattern information from a feature extracting circuit 2, the feature of a category to be recognized and its relating category string which are to be compared are inputted from a dictionary storage part 4. When these categories exist, their similarity is calculated on the basis of the prescribed algorithm. When the similarity is the prescribed value or more, the category concerned is outputted as proposed recognized result and its similarity is outputted. Whether the relating categories are registered in the category concerned or not is checked, and in case of existence, the relating categories are outputted as proposed recognized results and the similarity between the input pattern and the relating categories is also outputted.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an information recognition device that recognizes patterns by inputting pattern information such as characters, symbols, figures, etc., for example.

[Prior Art] In addition to keyboard input, a method of recognizing and inputting patterns of characters, symbols, figures, etc. has been considered as a simple method for inputting Japanese into information recognition devices such as computers and word processors. However, all of these input cover pattern information recognition devices recognize the input cover pattern information by comparing the input cover pattern information with recognition target categories registered in the pattern dictionary, so the input cover pattern information is not registered in the pattern dictionary. If the pattern is significantly different from the standard pattern, it may not be recognized or may be misrecognized.

For example, the characters shown in Figures 2 (a) to (f) are examples of lowercase English letters and symbols, but these patterns of lowercase English letters and symbols are particularly strongly influenced by the scribe's habits. , the fluctuations are also large. For example, when writing lowercase alphabento letters, some people use print (for example, Figure 2 (a) to (C)
)), while some people prefer to write in cursive (for example, characters other than the leftmost characters in Figure 2 (a) to (C)).
Some people only write.

In order to deal with such cases, it would be ideal to register patterns that assume all kinds of ancient characters and "broken" characters, but as a practical matter, this is close to impossible and is possible. However, it requires special recognition processing that is different from that for other character types, which poses a problem in terms of cost. Therefore, in conventional information recognition devices, the following measures have been taken for characters that vary widely depending on the scribe, such as lowercase English letters and symbols. In other words, such characters are not recognized as characters to be recognized, but are instead inputted directly as codes using Bentouch or a keyboard, or even when recognition is performed, the scribe is subject to fairly strict writing restrictions (e.g., inputting only lowercase letters in print). input recognition). However, it goes without saying that these countermeasures impose an unnecessary operational burden on the operator.

[Problems to be Solved by the Invention] The above is based on the fact that characters that have a scribe's quirks are also difficult for recognition devices to recognize. Therefore, an object of the present invention is to provide an information recognition device that does not require any special recognition processing, or in other words, has only a simple recognition method, but can still be used for information recognition that may vary depending on the following, such as lowercase letters or markings. An object of the present invention is to provide an information recognition device that enables input of large characters.

[Means for Solving the Problems] In order to solve the above problems, the information recognition apparatus of the embodiment shown in FIG. 1, for example, recognizes characters, symbols, etc.

An information input device 1 that inputs pattern information such as figures, a feature extraction circuit 2 that extracts features of manually generated pattern information, and features of various input pattern recognition target category strings 16 and association in advance with the recognition target category list 16. The dictionary storage circuit 4 stores and registers the 17 columns of related categories that have been stored, and the similarity between the input pattern features extracted by the feature extraction circuit 2 and the features of each recognition target category 16 stored and registered in the dictionary storage circuit 4. and an identification circuit 3 that calculates and selects categories 17 having a similarity of a predetermined degree or more as candidates for recognition results.

[Function] In the ``-'' structure acid, for example, when ``('' is input to the information input device 1, the identification circuit 3 selects the related categories (<, [, ir, r, c, etc.) in addition to ``(''.

-) is also output as a recognition result candidate. Therefore, the operator does not have to take the trouble of manually creating complex symbols such as (, c, (, r, r, (, O=), but only manually creates "(", which is morphologically similar to these symbols. In this way, the same effect as when manually inputting the F symbol to the information recognition device can be obtained.Also, for the information magazine a device, it is only necessary to recognize “(”, which is easy to recognize as a pattern. The algorithm is also simple.

[Example] Hereinafter, the example will be described in more detail with reference to the accompanying drawings.

Basic example FIG. 1 is a diagram showing the basic configuration of an embodiment of the present invention.

Reference numeral 1 denotes an information input device, such as a data tablet, a combination of a display and a light pen, or a device 6 for optically reading character patterns. Reference numeral 2 denotes a feature extraction circuit, which extracts features of the input pattern information inputted from the information input device 1 and sends them to the identification circuit 3. The identification circuit 3 calculates the similarity by comparing the features of the input pattern obtained from the feature extraction circuit 2 with the features of the standard pattern of the recognition target category registered in the dictionary storage circuit 4, and calculates the similarity when the similarity is determined to be a predetermined value. This circuit outputs the category when the value is greater than or equal to the value. Reference numeral 4 denotes a dictionary storage circuit that stores features of standard patterns of categories to be recognized. Reference numeral 5 denotes a control circuit, which controls the entire embodiment, and includes a CPU (central processing unit) 6, a ROM (read only memory) 7, and the like. A control procedure program as shown in the flowchart of FIG. 6 is stored.

<Features of Basic Embodiment> The features of this embodiment will be described below. In this embodiment, in the dictionary storage circuit 4, the characteristics of the target Q pattern and related categories related to the category are registered in association with the category to be recognized. Therefore, the identification circuit 3 uses the characteristics of the standard pattern of the recognition target category and the input pattern 6? When the degree of similarity with the N features is greater than or equal to a predetermined value, not only the category but also related categories related to the category are output. Differences in the basic concept between this embodiment and the conventional example are shown in FIGS. 4(a) and 4(b). The relationship between the input pattern lOO and the standard pattern (110, 111° 112) in the conventional example shown in FIG.
to, iit, t12) are all equal. On the other hand, the fourth
In this embodiment shown in Figure (a), each of the standard categories (101, 102, 103) corresponds to the input cover pattern 100, and related categories (101a, 1o1b, 102a, 102b, 103
a, 103b) correspond in a tree structure. In the conventional example, the judgment was made based on whether the patterns were similar or not, but in this embodiment, related categories that are not likely to be judged to be directly similar to the input cover pattern are also output as candidates. By doing this, when inputting characters such as lowercase letters, where the scribe has a strong habit of writing, the scribe only needs to input the standard category, and the standard category is easy for the recognition device to recognize and prevents recognition errors. do not have. Therefore, related categories related to this standard category are also output as recognition candidates, and the scribe selects a desired one from the related categories. The effect of Ic can be understood from the fact that when inputting the special symbols shown in the related categories in FIG. 5, it is only necessary to input symbols belonging to simple standard categories manually.

By the way, the degree of similarity mentioned above may be the degree of similarity of the category compared with the characteristics of the input cover pattern information, or it may be a value obtained by subtracting a predetermined value from this degree of similarity, or it may be a value obtained by subtracting a predetermined value from this degree of similarity. It is also possible to use the same value.

Operation of Basic Embodiment> The operation of the identification circuit 3 having the above characteristics will be explained with reference to the flowchart of FIG. The flow starts when the identification circuit 3 obtains the features of the input pattern information from the feature extraction circuit 2. First, in step S2, the characteristics of the recognition target category to be compared and the related category string are input from the dictionary storage circuit 4. In step S2, if such a category does not exist, the identification circuit 3 ends its operation. On the other hand, if such a category exists, the process proceeds to step S4, where the degree of similarity is calculated using a predetermined algorithm. Specifically, the predetermined algorithm may be, for example, mean square error 1 variance, entropy, etc. In step S6, whether the degree of similarity is greater than or equal to a predetermined value,
That is, it is checked whether there is a standard pattern that is similar to the input pattern to a certain extent or more. If the degree of similarity is not equal to or greater than the predetermined value in step S6, this category is rejected and the process returns to step Sl.

If the similarity is greater than or equal to the predetermined value in step S6,
Proceeding to step S8, the category is output as a recognition result candidate, and its similarity is output. Next, the process advances to step SIO, and it is checked whether a related category is registered for the category in question. This related category has been input to the identification circuit 3 in step S1. If there are no related categories, return to step Sl and repeat the steps described above until there are no more categories to compare in step S2. On the other hand, if a related category exists, the related category is output as a candidate for the recognition result, and the degree of similarity between the input cover pattern and the related category is also output. In this way, not only categories similar to the input cover pattern but also related categories related to that category are output! It becomes 1 yen.

<Example of related categories> Figure 5 is an example of related categories for each category.
" is a related category. The same goes for "B" and "C". It can be said that these are categories with similar "ideas."

On the other hand, in examples of relationships that are similar in form, °゛('' is replaced by
""c""-" is the related category, and for ")", ">""]"'l""J""j"]′"′ψ"
is a related category.

For ゛°zero'', ゛(5” ” @ ” ” & ” −★pa is the related category.

For example, when a pattern similar to ゛(' is input, the identification circuit 3 in FIG. When it is greater than or equal to a predetermined value, °
” “[” “I” “r” “r” “[”-
” is also output as a recognition result candidate.

Modified example> By the way, the related categories shown in FIG. 5 have priorities in the order shown in the diagram. Therefore, FIG. 6 shows an embodiment having a function to change the priority order. is written in the information input bag 511.

In addition to an input means 11 such as a tab left for inputting symbols 2 figures, etc., a function specifying means 12 is provided (
(The figure shows a case where a function to raise the priority of lowercase letters is specified), and the identification circuit 3 changes the priority of the recognition result based on the degree of similarity and the function designation. By the way, possible methods for specifying a function include pressing a predetermined key, touching a predetermined area, inputting a predetermined handwritten figure, and placing a mark in a predetermined area. The function processing circuit 14 is a circuit that sends a signal to the identification circuit 3 in accordance with a function specification specified by a predetermined means. A buffer 15 temporarily stores recognition result candidates and their similarities.

<Operation of Modified Example> The operation of the identification circuit 3 in the embodiment of FIG. 6 will be described with reference to the flowchart of FIG. 7. The operation of the identification circuit 3 is almost the same as that shown in the flowchart of FIG. If the different points are listed, step S8. Step S2
In step S12.6, decide whether to output the registered category and degree of similarity to the outside or to store it in a buffer. In step S30, decide whether to output the related category and degree of similarity to the outside or store it in a buffer; and step S2. The process differs in each point, such as whether to terminate when there is no category to be compared in step S21 or whether to execute steps 832 and subsequent steps.

Therefore, only step S32 in FIG. 7 will be explained below. In step S32, each category and its degree of similarity are read from the buffer 15, and those categories are arranged as candidate characters for the recognition result in 11 "1" with the highest similarity Jff.

In step S34, it is determined whether or not there is a function specification from the function processing circuit 14. If there is a function specification, candidate characters are rearranged according to the function specification and similarity in step 536, and candidate characters are output in order in step 338. do.

On the other hand, if there is no function designation in step S34, candidate characters are output in the same order. For example, in the embodiment shown in FIG.
When ``°A'' is input in 1, ``A'' and ``a'' are candidates for the recognition result, but due to the function specification of ``lowercase English letter'', ``°a'' has a higher priority, and `` ″
"a" becomes the first candidate. By pressing this, the next candidate specifying means 13 causes the identification circuit 3 to output the recognition result candidates in priority order.

In this way, by setting characters with poor recognition rate, such as lowercase English letters and symbols, to categories related to other categories, the user can input them in a simple manner. For example, to input lowercase English letters, specify the ``Lowercase English Letters'' function and then write down the corresponding uppercase letters.Also, to input symbols, enter a predetermined simple symbol, and the desired symbol will be displayed. All you have to do is specify the ``next candidate'' function until the next candidate is reached.

By the way, the above embodiment is an example in which the priority order can be changed by specifying a function. Therefore, it is also possible to automatically change the priority order for outputting the above-mentioned related categories using a learning function. This is identification circuit 3
The number of times the operator has selected the related categories outputted by is stored for each related category, and the priority order is changed in descending order of the number of selections. According to this method, the priority order is automatically changed, improving operability.

[Effects of the Invention] As described above, according to the present invention, special recognition processing is not required, and characters and symbols that vary greatly depending on the scribe, such as lowercase English letters and symbols, can be recognized.

It is possible to do it manually by drawing figures etc.

[Brief explanation of drawings]

FIG. 1 is a basic configuration diagram of an embodiment according to the present invention, and FIG.
a) to (f) are graphical diagrams showing examples of English lowercase letters, symbols, etc. that are difficult to recognize; FIG. 3 is a flowchart showing an outline of the control operation according to the embodiment; FIGS. 4(a) and (b) are FIG. 5 is a diagram illustrating the relational structure of categories between the embodiment of the present invention and the conventional example. FIG. 5 is an example diagram showing the correspondence between categories and related categories in the embodiment. FIG. 6 is another embodiment of the present invention. The configuration diagram of the example, FIG. 7, is a flowchart for explaining the operation of the recognition circuit of the embodiment shown in FIG. In the figure, l...information human powered device, 2...feature extraction circuit,
3... Identification circuit, 4... Dictionary tα circuit, 11...
・Character input area, 12...Parental lowercase letter function designation means, 13...'Next candidate function designation means, 14.Function processing circuit, 15.
・Buffer, 16... Recognition target category string, 17.
...is a related category column. Patent applicant Canon Co., Ltd. (0) l), necessary, b (b) e, buJ (d) (, (. (e) l also. (f) odd ★ Figure 3 Figure 4 (0) Figure 4vA(b) +10 11+ 112th 58th factor

Claims (5)

[Claims] (1) An input means for inputting pattern information such as characters, symbols, figures, etc., a feature extraction means for extracting features of the input pattern information, features of a recognition target category and related categories previously associated with the recognition target category. and a dictionary storage means that stores and registers the information, and calculates the degree of similarity between the features of the input pattern extracted by the feature extraction means and the features of each recognition target category stored and registered in the dictionary storage means, and the degree of similarity is equal to or higher than a predetermined degree. an identification unit that uses categories having a predetermined degree of similarity as candidates for recognition results, and the identification unit uses the related categories as candidates for recognition results in addition to the categories having the predetermined degree of similarity. Device. (2) The information recognition device according to claim 1, wherein the relationship between the recognition target category and the related category is a relationship between uppercase and lowercase letters. (3) The information recognition device according to claim 1, wherein the relation between the recognition target category and the related category is a morphologically similar symbol or character. (4) Related categories corresponding to each recognition target category are stored and registered in the dictionary storage means in order of priority, and become candidates for recognition results in the order of the priority. The information recognition device according to any of paragraph 3. (5) The information recognition device according to claim 4, further comprising priority order changing means for changing the priority order.