JPS6120036B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a highly practical character reading device that can read a large number of characters, including Chinese characters, and stably and efficiently output features for major classification identification.

Technical Background of the Invention The characteristics of an input character pattern are extracted and compared with the characteristics of dictionary patterns registered in advance in a dictionary, and the input character pattern is read and recognized according to the results of the comparison. . However, when a large number of characters including Chinese characters are to be read in this character recognition, the amount of processing required to match the above-mentioned features becomes extremely large, and the processing efficiency deteriorates significantly.
Therefore, conventionally, prior to the above recognition process, characters to be recognized are roughly classified based on a plurality of general characteristics.
This reduces the number of characters to be compared in the identification stage, thereby improving processing efficiency and speeding up the processing speed. At this time, it is necessary to use the same features as the features used for the above-mentioned major classification and the features used for candidate character identification, but this feature is necessary to prevent deformation of character patterns caused by handwritten characters and various noises. It needs to be sufficiently stable.

However, conventionally, the features used to broadly classify character patterns have focused on the complexity of the character lines, the shape of the peripheral part of the character, and even the rough shape of the entire character. There are things that have been done. In addition, these are introduced in detail in, for example, the following literature.

Sakai, Watanabe “Current Status of Printed Kanji Recognition” Information Processing, Vol. 22 No. 4 PP. 274-279 (April 1981) Figure 1 shows an example. Focusing on the shape around the original character pattern, The character lines existing in the rectangular scanning areas U, D, L, and R on the upper, lower, left, and right sides of the surrounding area are detected, and the amount is set to "0", "1",
It is quantized into three levels of "2" and used as feature data. In addition, the one shown in Fig. 2a is the circumscribed edges u, d, which touch the original character pattern,
Focusing on l, r, these circumscribed edges u, d, l, r
The area of the character background that is in contact with is determined and used as a multidimensional feature vector. The difference is that the former (Figure 1) focuses on the character line area, while the latter (Figure 2 a) focuses on the area of the background area, but both of them focus on the area quantity around the character pattern. It can be said that it is a characteristic.

Problems with the Background Art However, the above feature extraction is very effective in cases where the pattern is standardized, such as with printed characters, but when the pattern is standardized, such as with handwritten characters, even though the pattern has characteristics as a character, If it is significantly deformed, it is extremely unstable. Moreover, even if the character is a printed character, if the character pattern is chipped or faded as shown in FIG. 2b, the feature extraction becomes extremely unstable. In other words, there is a problem in that feature extraction cannot always be performed stably against deformation of character patterns and various noises.

Purpose of the Invention The present invention has been made in consideration of the above circumstances, and its purpose is to broadly classify input original character patterns by reading a large number of characters including handwritten kanji. It is an object of the present invention to provide a highly practical character reading device that can stably and efficiently extract the features necessary for the text.

Summary of the Invention The present invention focuses on the fact that a large amount of classification information (features) necessary for large classification identification of characters exists in the periphery of a character pattern, and calculates feature quantities in a wider area around the character pattern. , it is possible to use relative position information instead of absolute position information and to use the positional relationship between features as a new feature. The position of the center of gravity of the background toward the center is determined for each circumscribing edge, and the average amount of indentation of the character pattern is determined from the relative position of these center of gravity positions with respect to the circumscribed frame, and this is extracted as a feature for major classification identification. It is in a character reading device that has been designed.

Effects of the Invention Therefore, according to the present invention, a novel concept of center-of-gravity position information of a background portion corresponding to each circumscribed side of the character pattern surrounded by circumscribed sides touching the input character pattern is introduced, Moreover, since the feature is the relative positional information of these centroid positions with respect to the circumscribed frame, it is possible to obtain the feature information with sufficient stability against deformation of character patterns and various types of noise. Therefore, even if the printed characters are chipped or faded, or even if they are handwritten, the characteristics necessary for broad classification identification can be obtained stably and efficiently, making it easier for character recognition processing. Significant improvements can be made. In addition, even when dealing with complex character patterns such as kanji and handwritten characters, information that strongly reflects the characteristics of the original character pattern is obtained from the center of gravity of the background and its relative position. It is possible to achieve a highly practical and remarkable effect in character recognition processing, such as being able to easily obtain .

Embodiment of the Invention Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

FIG. 3 shows the feature extraction process of the input character pattern in this apparatus, and here the character pattern "上" input by hand is shown. The feature extraction process is performed by first detecting a circumscribing frame surrounding a given character pattern. This circumscribing frame has a left circumscribing edge L that touches the leftmost edge of the character pattern, a right circumscribing edge R that also touches the rightmost edge, an upper circumscribing edge U that touches the uppermost edge of the character pattern, and a lower circumscribing edge D that touches the lowermost edge. It is determined by detecting. The circumscribed sides U, D, L, and R can be easily determined by scanning and detecting the character pattern and comparing the position coordinates of the character lines.

After that, each of the obtained circumscribed sides U, D, L,
With R as the reference position (reference line), each circumscribed side U,
Scanning is performed from D, L, and R in the direction toward the center of the character pattern, and the length of the scanning line until reaching the character line is determined. However, if the maximum length of the above scan is set to, for example, the length l
It is preferable in terms of signal processing to specify the value as (variable) depending on the size of the character pattern. Then, the area S of the background portion of the character pattern corresponding to each of the circumscribed sides U, D, L, and R indicated by the sum of these scanning line lengths is determined. For example, the area S of the background part on the circumscribed side U is determined by scanning in the direction toward the center of the character pattern with the circumscribed side U as the reference position using a scanning line defined to the maximum length l as shown in Fig. seek. However, when the scanning line reaches the character line, the scanning ends, and if it does not reach the character line, the scanning line ends when the length l is reached. Then, assuming that each point of the area indicated by the area S obtained in this way is (x, y), the center of gravity position G(X, y) of the area S is
Y) is determined as follows.

X=〓 _s x. S _(x.y) d _{x dy} /〓 _s S _(x.y
) d _x d _y Y=〓 _s y. S _(x.y) d _{x dy} /〓 _s S _{(x.y
) d _} The information reflects the characteristics of Accordingly, the centroid positions of the background portions touching the other circumscribed sides D, L, and R, respectively, are determined by similar processing.

Then, from the information on these centroid positions, the relative position Pu of the centroid G of the area S normalized for the size of the circumscribing frame of the character pattern, that is, the vertical and horizontal widths, with respect to the character frame is calculated as P _u = (X/H, Y/V). However, H and V in the above formula indicate the horizontal width and vertical width of the character pattern circumscribing frame. This normalization process is performed for each of the centroid positions G of the background portion that are in contact with each of the circumscribed sides U, D, L, and R. As a result, information indicating the characteristics of the surrounding area of the input character pattern is converted to the normalized center of gravity position {P _U , P _D , P
_L , P _R }. This information corresponds to the average amount of indentation of the outer shape seen from the circumscribed sides U, D, L, and R of the character pattern, and therefore, the characteristics of the character pattern can be used to roughly classify and identify character patterns. This is a sufficient reflection. Therefore, by using the characteristics of the surrounding area of the character pattern obtained in this way {P _U , P _D , _PL , P _R }, the character pattern can be effectively, stably and reliably classified into major categories. It becomes possible to do so.

For example, if the characters ``ku'' and ``e'', which have different sizes and are handwritten, are input as shown in Figure 4 a and b, respectively, then the characteristics of the surrounding area of the character pattern can be determined by the process described above. It is shown as {P ¹ _U , P ¹ _D , P ¹ _L , P ¹ _R } {P ² _U , P ² _D , P ² _L , P ² _R }. Therefore, if these characteristic information are superimposed and compared as shown in FIG. 4c, it is clear that there is a remarkable difference in the average amount of depression. In particular, if we compare features that correspond to the same circumscribed edge, for example the distance between the normalized centroid positions, we can see that the difference in features largely reflects the difference in character pattern features as the size of the distance. I understand. Therefore, regarding the similarity between the two, the distances (P ¹ _U , P ² _U ), (P ¹ _D , P ² _D ), (P ¹ _L , P ² _L )
If each of (P ¹ _R , P ² _R ) is determined, a comprehensive judgment can be made, and it is possible to easily identify major classifications. Furthermore, the above determination can be effectively performed through arithmetic processing.

FIG. 5 is a schematic configuration diagram showing an embodiment of an apparatus of the present invention which performs such processing to extract features for broad classification identification of character patterns.

A character pattern 1 to be read is read by a scanning photoelectric conversion device 2 consisting of, for example, a television camera.
The data is photoelectrically converted and input by scanning the character surface. The image signal of the character pattern 1 inputted via the photoelectric conversion device 2 is inputted to the binary quantization device 3, where it is discriminated at a discrimination level determined based on, for example, the background density and is binary quantized. . The quantized character pattern signal is then stored in a pattern storage device 4 such as a one-frame memory as a binary pixel signal corresponding to the scanning position. The character pattern pixel signals stored in the storage device 4 are scanned in a predetermined direction and read out, and subjected to the feature extraction process described above.

That is, the circumscribing frame detection device 5 searches for the positional coordinates of the character line of the character pattern stored in the storage device 4, and determines its leftmost coordinate X _L and its rightmost coordinate X _R
The left circumscribed side L is determined as L=X _L , and the right circumscribed side R is determined as R=X _R. At the same time, the uppermost end coordinate Y _U and the lowermost end coordinate Y _D of the character line of the character pattern are determined, and the upper circumscribed side U and lower circumscribed side D are determined. Then, the circumscribing frame detection device 5 sends information about these circumscribing frame sides L, R, U, and D to scanning circuits 6 _L , 6 _R ,
6 _U and 6 _D are given as control information, respectively.
In each of the scanning circuits 6 _L 6 _R , 6 _U , and 6 _D , a reference line for starting scanning is determined based on the information on the given circumscribed side, and the signal of the character pattern stored in the storage device 4 is transferred to the reference line. In the direction from to the center of the character pattern, (left → right), (right → left), (top → bottom), (bottom →
(above) and calculates the sum of the scanning line lengths corresponding to (contacting) each circumscribed side. In other words, the background area S is calculated using each circumscribed side as a reference. The center of gravity calculation units 7 _L , 7 _R , 7 _U , and 7 _D that obtain information on the area S obtained in this way calculate the center of gravity positions G (XY), respectively, as described above, normalize these, and relative The normalized center of gravity positions P _L , P _R , P _U , and _PD , which are general position information, are obtained respectively.
These pieces of information are summarized and given to the feature comparison device 8 as feature information { _PL , _PR , _PU , _PD } of the surrounding area of the input character pattern.

In the feature dictionary 9, the above-mentioned feature information for each character is registered in advance, and the feature comparison device 8 calculates the similarity (degree of similarity) between these dictionary features and the obtained features of the input character pattern. Calculated sequentially. Then, when the degree of similarity is greater than or equal to a predetermined tolerance value θ, the category of the obtained feature dictionary is output as a major classification identification result.

In this way, the embodiment device that performs feature extraction and broad classification identification based on the feature extraction according to the present invention can be realized very easily. Then, by feeding the thus obtained large classification identification results and the image signal of the input character pattern to the character recognition section at the next stage, the character recognition can be performed easily and efficiently. Moreover, as described above, it is stable against deformation of character patterns such as handwritten characters and various noises, so that the efficiency of recognition processing can be significantly improved. Therefore, it is possible to construct a highly practical character reading recognition system that recognizes many characters including Chinese characters, and has great effects.

Note that the present invention is not limited only to the above-described embodiments. For example, in the embodiment, the center of gravity of the region S was obtained as the feature information, but to obtain a similar effect, the center of the region S was determined as X=〓 _s S _(x.y) d _{x dy} /H・V Y= It may be obtained as _s S _{(x, y)} d _{x dy} /H·V, and this may be used as the feature information. In this way, data processing and device configuration can be simplified. Furthermore, feature information of character lines may also be used for broad classification identification. In short, the present invention can be implemented with various modifications without departing from the gist thereof.

[Brief explanation of the drawing]

1 and 2 a and b are diagrams for explaining the concept of conventional character pattern feature extraction; FIG. 3 is a diagram for explaining the concept of character pattern feature extraction according to the present invention; 4a to 4c are diagrams for explaining the differences in character patterns indicated by the characteristics of character patterns according to the present invention, and FIG. 5 is a schematic diagram of the main part of an apparatus according to an embodiment of the present invention. 1... Character pattern, 2... Scanning photoelectric conversion device, 3... Binary quantization device, 4... Pattern storage device, 5... Circumscribing frame detection device, 6 _L , 6 _R , 6 _U ,
6 _D ...Scanning circuit, 7 _L , 7 _R , 7 _U , 7 _D ... Center of gravity calculation unit, 8... Feature comparison device, 9... Feature dictionary.

Claims (1)

[Scope of Claims] 1. Means for photoelectrically converting and inputting the original character pattern, means for binary quantizing and storing the input original character pattern, and 4-direction conversion of the stored original character pattern in four directions: top, bottom, left and right. means for detecting a circumscribed frame consisting of circumscribed edges; means for detecting the center of gravity or center position of a character background portion in a direction from each of the circumscribed edges toward the center of the original character pattern; A character reading device comprising means for broadly classifying and identifying the original character pattern according to the relative position of the center of gravity or center position of the character background portion with respect to the circumscribing frame. 2. The means for detecting the center of gravity or center position of the character background area is to scan the inside of the circumscribing frame from each circumscribing edge toward the center of the original character pattern, and to calculate the scanning distance from each circumscribing edge to the character line. A character reading device according to claim 1, wherein the character reading device is made by: . 3. The character reading device according to claim 1, wherein the relative position of the center of gravity or center position of the character background corresponding to each circumscribed side with respect to the circumscribed frame is determined as normalized position data. 4. Broad classification identification of the original character pattern is performed using the data of each set of relative positions with respect to the center of gravity or the central circumscribing frame of the character background corresponding to each circumscribing edge as characteristic information of the surrounding area of the original character pattern. A character reading device according to claim 1.