JPH04167193A - Character recognizing method - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a character recognition method for reading characters printed on paper or the like as images and recognizing each character.

[Prior Art] Conventionally, when recognizing characters from a character pattern, a document to be recognized is placed on an image reading device, an image of the document is read as a bit image, and each character is cut out from the document. Then, while comparing the feature amount of the character pattern of each character with the feature amount of a previously stored character, if a match is found, that character is output as a recognition result.

[Problems to be Solved by the Invention] However, in the conventional case, the characters cannot be recognized unless the original is placed in the specified direction (top, bottom, left and right) or the original is placed in the specified direction. It was necessary for the user to instruct the translation device about the direction of placement. Or,
The processing speed was slow because the dictionary was checked in all four directions.

[Means for Solving the Problem] In order to solve this problem, the present invention focuses on the position of punctuation marks to identify the direction of characters.

In other words, it is determined whether the recognition target document is written vertically or horizontally, and based on the size and position of the circumscribed rectangle of the characters constituting the recognition target document, the circumscribed rectangle of the character is determined if the character exists in a position corresponding to a punctuation mark. The system calculates the punctuation mark area, identifies the direction of the character according to the position of the punctuation mark within that area, and performs character recognition.

[Operation] Punctuation marks are located at the bottom left of the area in horizontal writing, and at the top right in vertical writing, so the direction of the characters can be specified using the information as to whether the document is written vertically or horizontally.

[Embodiment] A character recognition method according to an embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram of a character recognition device using the character recognition method of this embodiment. 1 is a photoelectric conversion unit that optically scans a written document and converts it into an electrical signal; 2 is a photoelectric conversion unit that converts the photoelectric conversion signal sent from photoelectric conversion unit 1 into black and white according to a certain standard; A binarization unit performs binarization, 3 is an area division unit that divides the binarization pattern sent from the binarization unit 2 into areas, and 4 is a character area extraction unit that extracts character areas from the divided areas. , 5 is a vertical writing/horizontal writing determination unit that creates a horizontal/vertical projection of the character area extracted by the character area extraction unit 4 and determines vertical writing/horizontal writing from the interval, 6
is an orientation determination unit that determines the orientation of the character area extracted by the character area extraction unit 4. Reference numeral 7 denotes a character recognition unit that performs character recognition according to the results determined by the vertical/horizontal writing determination unit 5 and the orientation determination unit 6.

FIG. 2 is a block diagram of the character recognition device of FIG. 1.

A RAM 8 stores image data and the like. A ROM 9 stores a character pattern dictionary, programs, and the like. 10 is R3232C, which issues recognition commands and outputs characters. 11
is an arithmetic processing unit that controls the operation of a program. Reference numeral 12 captures an image into an image memory using a scanner.

The flow of the entire process can be explained using the flowchart in Figure 3.
explain.

First, binary data is scanned in both the vertical and horizontal directions and divided into regions using continuous portions of black and white pixels as region dividing lines (step 1, FIGS. 9 and 10). The divided area is further divided into small areas of 8 connected black pixels, and the interval between them is
A region with little variation in size distribution is extracted as a character region (Step 2, FIG. 11). The horizontal and vertical oblique shadows of the extracted character area are taken, and the vertical or horizontal writing is determined based on the size of the valley (step 3, FIG. 4).

Punctuation mark candidates are extracted from the position and size of the small area of 8 connected black pixels and recognized (step 4, FIG. 5). As a result of the recognition, if it is determined that it is a punctuation mark based on the evaluation value such as its similarity, the punctuation mark area that is the circumscribed rectangle of the character if a character exists corresponding to the position of the punctuation mark is calculated, and the punctuation mark area is calculated. The orientation is determined based on the position of the punctuation mark within the punctuation mark area. Step 8) If all punctuation marks are rejected, four patterns are created by rotating several regular character patterns by 90 degrees, 180 degrees, and 270 degrees. recognition is performed (step 6). For these four types, the orientation is determined based on the direction in which the sum of recognition accuracy such as similarity is the largest (step 7). Finally, all characters are recognized assuming that the target character area has the determined vertical writing/horizontal writing and orientation (step 9).

The explanation of orientation determination will be supplemented with a specific example. If the punctuation mark is at the bottom left as shown in Figure 8 (a), if it is written horizontally, it will be in Figure 6 (a) portrait, upright; if it is written vertically, it will be (d) portrait, upside down (rotated 180 degrees). It can be determined that Furthermore, if the characters are rotated by 90 degrees as shown in FIG. 7(b), it can be determined that the characters are in FIG. 6(f) landscape, reverse orientation, and FIG. 6(h) landscape, normal orientation.

[Effects of the Invention] As described above, the present invention makes it possible to perform character recognition fully automatically by determining the orientation of a manuscript from information about vertical writing and horizontal writing and the position of punctuation marks, thereby saving labor. This has the effect of eliminating operational errors.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of a character recognition device using a character recognition method according to an embodiment of the present invention, FIG. 2 is a block diagram of a character recognition device according to an embodiment of the present invention, and FIG. 3 is a block diagram of a character recognition device according to an embodiment of the present invention. Flowchart, FIG. 4 is a diagram showing an example of horizontal and vertical oblique shading, FIG. 5 is an example of a circumscribed rectangle of 8 adjacent small areas, FIG. 6 is a diagram showing an example of vertical writing, horizontal writing, and orientation.
Figure 7 is a diagram showing an example of a character pattern coming in from a scanner, Figure 8 is a diagram also showing an example of punctuation marks, Figure 9 is a diagram showing an example of an input document, and Figure 10 is a diagram showing an example of an input document. FIG. 11 is a diagram showing an example of region division of an input document.
It is a figure which shows the example which divided|segmented into connected small connections and extracted the character area. Figure 2 Figure 4 Figure 5 Figure 7 (a) (b) (c) (d) Figure 8 (a) (b) (c) (d) (○) (b) (e) (f ) Figure (c) (d) (CI) (it) Figure 9 Figure 11 Figure 10

Claims (1)

[Claims] 1) Read an image of a document to be recognized, find a circumscribing rectangle of each character in the read image, and match punctuation marks in the read image based on the size and arrangement of the obtained circumscribed rectangle. If a character exists at that position, find a punctuation mark area that is a circumscribed rectangle of the character, and identify the orientation of the document to be recognized according to the position of the punctuation mark within the punctuation mark area and information on whether it is written vertically or horizontally. , A character recognition method characterized in that each character in a read image is recognized according to the identified orientation. 2) The punctuation mark area is determined for a plurality of punctuation marks, and characters are recognized according to the most common direction among the directions of the document obtained from each of the punctuation marks. Character recognition method described.