JPH02195430A - Character segmenting circuit - Google Patents

Abstract

PURPOSE:To separate overlapped characters which can not be separated from each other by a projection method and to segment the separated characters by providing the circuit with a segmenting means for segmenting a character pattern by separating position information obtained in each scanning line. CONSTITUTION:An address generating circuit 21 in a character segmenting circuit is connected to a memory 11 through a bus 21 and an address for partially scanning an input image stored in a memory 11 is generated. A changing point detecting part 31 is connected to memories 11, 12 through buses 71, 72, a point for changing data read out from the memory 11 from black to while and from white to black is detected and positional information in each white block obtained from the address of the changing point is stored in the memory 12. A structure extracting part 41 is connected to the memories 12, 13 through the buses 72, 73, the positional information of the white block is read out from the memory 12 and contact information between respective white blocks is extracted by deciding the existence of contact between respective white blocks and stored in the memory 13.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of the Invention The present invention relates to a character extraction circuit for a character recognition device, and in particular, to a character extraction circuit for a character recognition device.
Regarding character cutting technology.

BACKGROUND OF THE INVENTION As shown in FIG. 2, a character cutting circuit in a conventional character reading device extracts a projection pattern of an input binary image and calculates the effective,
Invalidity was determined, and the pattern in the range where valid black blocks existed was extracted as a character pattern.

Problems to be Solved by the Invention However, since the conventional character extraction circuit described above performs character extraction only from projection information, it has the disadvantage of not being able to separate overlapping characters or noise in the direction of projection. there were. .

For example, in the case of the input pattern shown in Figure 3(a), the character "\"
” and the character “4” overlap in the X direction, so the character “¥” and the character “4” cannot be separated due to the projection pattern.Also, the upper noise and the character “¥” overlap in the Y direction, so the projection pattern ■ The character “¥” and noise cannot be separated.

In such cases, some conventional character cutting circuits forcibly separate the character pattern at the position indicated by character frame position information given in advance, but in this case, a third As shown in Figure (b), character deformation occurs due to character cutting, and parts of adjacent characters and noise remain, which has the disadvantage of deteriorating the performance of character recognition performed after cutting.

The present invention has been made in view of the above-mentioned conventional situation,
SUMMARY OF THE INVENTION Accordingly, an object of the present invention is to provide a novel character cutting circuit which makes it possible to eliminate the above-mentioned disadvantages inherent in the conventional technology.

Means for Solving the Problems In order to achieve the above object, a character cutting circuit according to the present invention includes a memory capable of storing binary image information for one character or more;
means for partially scanning the input image stored in the memory and detecting positions where the scanned binary image changes from black to white and from white to black; means for storing positional information; means for reading out the positional information and determining whether or not there is contact between the respective blocks; and extracting contact information between the respective blocks; and sequentially contacting the white blocks on the first scanning line based on the contact information. means for searching for a white block and extracting a path that can reach the white block on the final scanning line; means for extracting separation position information for each scanning line for character separation from the path obtained by the extraction means; and means for cutting out a character pattern based on separation position information for each scanning line.

Embodiment Next, a preferred embodiment of the present invention will be specifically explained with reference to the drawings.

FIG. 1 is a schematic block diagram showing an embodiment of the present invention.

Referring to FIG. 1, one embodiment of the present invention includes a memory 11
．． 12.13.14.15, address generation section 21, change point detection section 31, structure extraction section 41. It is composed of a structure analysis section 51, a pattern cutting section 61, and paths 71, 72, 73, 74, and 75.

The address generator 21 is connected to the memory 11 via a path 71 and generates an address for partially scanning an input image stored in the memory 11.

The change point detection unit 31 connects to the memory 1 via paths 71 and 72.
1.12, detects the point where the data successively output from the memory 11 changes from black to white and from white to black,
Position information for each white block obtained from the address of the change point is stored in the memory 12.

The structure extraction unit 41 accesses the memory 12 via paths 72 and 73.
．． 13, reads out the position information of the white block stored in the memory 12, extracts contact information between each block by determining whether there is contact between each block, and stores it in the memory 13. The contact judgment between
For example, there is contact with a white block that has a white point immediately above or below the white point of any of its own blocks, such as between ■ and 0 in Figure 4 (al) and between ■ and 0; ), a white block that does not have a white point above or below any white point is determined to have no contact. By reading out the contact information between each block stored in the memory 13, contact between each block can be determined. For example, by reading out the memory 13, it is possible to create a connection diagram in which each block is set as a node and each node is connected when there is contact, as shown in FIG. 4fa) and ibl. Figure 4(a) shows the case where there is contact between white blocks ■ and 0 and between white blocks ■ and 0, and Figure 4(b)
l is an example where there is no contact between ■, ■, ■, and 0.

The structure analysis unit 51 accesses the memory 13 via paths 73 and 74.
．． 14, and while reading the contact information between each block stored in the memory 13, searches for white blocks that come into contact sequentially from the white block on the first scanning line, and a path that can reach the white block on the final scanning line. Find out everything about. At this time, if there are two or more paths from the first scanning line to the final scanning line, the optimal path with the smallest maximum distance between the white point closest to the scanning center of each block and the scanning center, that is, the scanning center. On the other hand, the path with the smallest shake of the white block is determined. Furthermore, if there is no single path, that path is determined to be the optimal path. The positional information of the white point closest to the scanning center of the all-white block included in the optimal path is stored in the memory 14 as separation positional information for separating characters.

FIG. 5 shows an example of a scanning pattern and a separation position for the scanning pattern.

The pattern cutting unit 61 is connected to the memory 11, 14, and 15 via paths 71, 74, and 75, and performs extraction processing of the separation position information on upper, lower, left, and right characters indicated by the position information of the character frame given in advance. The separation position information obtained after performing the separation around the frame is read out from the memory 14, the image inside the separation position is cut out as a character image from the image stored in the memory 11, and the cut out character image is stored in the memory 15.

The input pattern, scanning area, obtained separation position, and cutting pattern when FIG. 3(a) is input to this circuit are shown in FIG. 6 (al, lbl, (c), and (d)), respectively.

As described in detail, according to the present invention, two or more characters
a memory capable of storing value image information; a means for partially scanning the input image stored in the memory and detecting a position where the scanned binary image changes from black to white and from white to black; means for storing positional information of the obtained white blocks; means for reading out the positional information, determining the presence or absence of contact between the respective blocks, and extracting contact information between the respective blocks; means for sequentially searching for contacting white blocks starting from the white block and extracting a path that can reach the white block on the final scanning line; and extracting separation position information for each scanning line for character separation obtained by the extraction means. By having the means and the means for cutting out a character pattern based on the separation position information for each scanning line, it is possible to obtain the effect that overlapping characters that cannot be separated by a projection method can be separated and the characters can be cut out. , characters that cannot be separated by the projection method as shown in FIG. 3(a) can be completely separated and cut out using the circuit of the present invention as shown in FIG.
This has the effect of improving the performance of character recognition performed after extraction.

Furthermore, when the circuit of the present invention is used in an optical character reading device, since overlapping between characters is allowed, the spacing between characters written on a form can be reduced, and more characters can be written on a form of the same size. This has the effect of allowing you to write down the information.

[Brief explanation of the drawing]

FIG. 1 is a schematic block diagram showing an embodiment of the present invention;
FIGS. 2 and 3 are explanatory diagrams of the operation of a conventional character cutting circuit,
FIG. 4 is an explanatory diagram regarding contact determination between white blocks in a circuit according to an embodiment of the present invention, FIG. 5 is an explanatory diagram regarding separation position determination in a circuit according to an embodiment of the present invention, and FIG. 6 is an explanatory diagram according to the present invention. FIG. 2 is an explanatory diagram regarding the operation of a circuit according to an embodiment. 11.12, 13.14, 15...Memory, 21...
・Address generation section, 31... Change point detection section, 41...
・Structure extraction section, 51... Structural analysis section, 61... Pattern cutting section, 71, 72.73, 74.75... Pass patent applicant NEC Co., Ltd. Agent Patent attorney Yutabe Kumagai Figure 1 Connection area 1, 4 6 points [ko = ≠] White block diagram

Claims (1)

[Claims] In the character cutting circuit of a character reading device, there is a memory that can store binary image information for one character or more, and an input image stored in the memory is partially scanned, and the scanned binary image is changed from black to white and white. means for detecting a position where the color changes from to black;
means for storing positional information of consecutive white points (hereinafter referred to as white blocks) for each scanning line obtained by the detection means; and means for reading out the positional information and determining whether or not there is contact between each white block; means for extracting contact information of
Means for searching sequentially contacting white blocks from the white block on the first scanning line based on the contact information and extracting a path that can reach the white block on the final scanning line, and for separating characters from the path obtained by the extraction means. 1. A character cutting circuit comprising: means for extracting separation position information for each scanning line; and means for cutting out a character pattern based on the separation position information for each scanning line.