JPH0578068B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION Field of Industrial Application The present invention relates to a character recognition device that recognizes printed characters and handwritten characters from newspapers, magazines, etc., and converts them into an amount of information such as a JIS code.

Conventional technology Conventional character recognition devices target documents with clear formats such as vertical writing, horizontal writing, line spacing, character spacing, etc., in other words, documents where the absolute position of the characters to be read on the paper is known in advance. I've been recognizing it. This places restrictions on the documents that can be targeted by character recognition devices, and to solve this problem, two-dimensional Fourier transform of the input image is used to detect the line spacing and the line direction, such as vertical or horizontal writing, is detected. extract,
Even for documents with unknown formats, methods have been used to determine the order of recognition candidate characters so that the meaning of the text can be understood. (For example, Hase and Hoshino, “Periodic characteristics of printed character strings,” IEICE (D), J65-D, 2,
PP.298-299) Problems to be Solved by the Invention However, the conventional technique of detecting the line direction of an input document by two-dimensional Fourier transform of the input image requires a large amount of calculation due to two-dimensional Fourier transform. The process takes time because of the method used.

The present invention has been made in view of these points, and uses a simple method to detect the line direction of an input image, and edits recognition candidate characters so that the meaning of the text can be understood even for documents with unknown formats. The purpose of the present invention is to provide a character recognition device that can perform the following functions.

Means for Solving the Problems In order to solve the above-mentioned problems, the present invention scans an input image in the vertical and horizontal directions to obtain a histogram of pixels forming a character part, and By comparing the average value of the character interval length in each direction, the line direction of the document can be easily extracted and recognition candidate characters can be edited.

Effects The present invention makes it possible to extract the line direction of a document whose format is unknown at high speed and edit recognition candidate characters so that the meaning of the sentence can be understood by using the above-mentioned technical means.

Embodiments Hereinafter, embodiments of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram of an embodiment of a character recognition device according to the present invention. Reference numeral 1 denotes an image input unit which scans an image including characters to be recognized, inputs the image as a binary signal, and stores it in the image memory 2. Reference numeral 3 denotes a line direction determination unit which scans the image memory 2 to determine whether the input image is written vertically or horizontally, and at the same time detects the line address. Reference numeral 4 denotes a character cutting unit, which scans the image memory 2 line by line using the line direction and line address detected by the line direction determination unit 3, and extracts a character image to be recognized character by character using projection of the image in line units. Cut out,
Detect the character address on the input image for each character.
5 is a recognition unit, which calculates feature quantities such as strokes of characters to be recognized cut out by the character cutout unit 4;
The characters are compared with the feature amounts of the characters registered in advance in the dictionary 6, and the most similar character is selected as the recognition candidate character.
Reference numeral 7 denotes an editing section, which uses the line direction and line address obtained by the line direction determination section 3 and the character address obtained from the character extraction section 4 to understand the meaning of the sentence from the recognition candidate characters extracted by the recognition section 5. Edit in order,
It is stored in the text memory 8 as a character code.

The operation of the character recognition device configured as described above will be explained using the input image P shown in FIG. 2 as an example.

The image P input from the image input section 1 is stored in the image memory 2 as binary data of 1 for text and 0 for white. First, the line direction determining unit 3 scans the input image P stored in the image memory 2 to obtain a vertical histogram H _v and a horizontal histogram H _h of pixels forming a character portion in the entire input image.
The result will be as shown in the figure. In order to separate the character part and the inter-character part, the histograms H _v and H _h are divided into the inter-character part where the histogram value is less than or equal to 0 pixels, and the character part with the number of pixels greater than 0 pixel, and the beginning of each part is Ask for address. y _s1 , y _s2 ... in Figure 2
...y _si ... and x _s1 , x _s2 ...x _si ... are the start addresses of the character parts, y _e1 , y _e2 ... y _ei and x _e1 , x _e2
...x _ei ... is the start address of the intercharacter portion.
From this address, the vertical character spacing length (y _si+1 −
Comparing the average value ( _si+1 − _ei ) of y _ei ) with the average value (si _{+1 − ei ) of the horizontal character spacing length (x si+1 −x ei )} , it is found that the vertical value is the same as the horizontal value. It is found that the line direction of the input image P is horizontal writing. Furthermore, since the line direction of the input image P has been determined to be horizontal writing, the start addresses of the character portion and the inter-character portion of the histogram H _v are y _s1 , y _e1 ... y _si , y _ei ...
... becomes the row address of the input image P.

Next, the character cutting section 4 extracts the line image L shown in FIG. 3 from the image memory 2 using the line address. Project the extracted row image L to create a histogram
When H ₁ is determined, it becomes as shown in Figure 3. Histogram H ₁ to 1 character horizontal address (z _s1 ,
z _e1 )...(z _si , z _ei )... are obtained, the character image to be recognized is cut out character by character, and the character address is determined by combining with the line address.

The rectangles R of the cut out characters to be recognized are sequentially input to the recognition unit 5. FIG. 4a shows the character "pine" cut out to be recognized. For each pixel of the cut out "pine tree", it is checked whether or not there are M or more (M is preset) consecutive pixels including the pixel of interest in the direction indicated by the arrow in FIG. 4b, and a direction code is set. Strokes are extracted by examining the connectivity of each pixel for each direction code, and feature quantities such as the number, position, and length of strokes are extracted. Figure 4a shows the stroke extraction results for the recognition target character ``pine''. The extracted feature amount is compared with the feature amount of characters registered in the dictionary 6, and the most similar character "pine" is selected as a recognition candidate character.

The recognition candidate characters of the recognition target characters cut out from the input image P sequentially extracted from the recognition unit 5 are selected by the editing unit 7 as “pine”, “shita”, “”, based on the fact that the input image P is written horizontally and the respective character addresses. electricity” “vessel”
...and line up the letters horizontally from top left to bottom right,
The text is edited so that its meaning can be understood and stored in the text memory 8 as a character code.

With the character recognition device configured as described above, the meaning of a sentence can be understood by cutting out characters to be recognized and editing candidate characters for recognition using the line direction and line address obtained using a simple method. You can create strings. Furthermore, by connecting the text memory to a document processing device or the like, new document editing becomes possible.

In this example, the histogram used to determine the row direction was obtained for the entire input image, but in order to deal with the inclination of the rows, the input image was divided into blocks and a histogram was obtained for each block. , the row direction can be determined.

Effects of the Invention According to the present invention, an input image is scanned in the vertical and horizontal directions to obtain a histogram of pixels forming a character portion, and the average values of the character spacing lengths in the vertical and horizontal directions obtained from the histogram are compared. By quickly extracting the line direction of an input image using a simple method, it becomes possible to edit recognition candidate characters in a document whose format is unknown so that the meaning of the document can be understood.

[Brief explanation of drawings]

FIG. 1 is a block diagram of a character recognition device according to an embodiment of the present invention, FIG. 2 is an explanatory diagram of an input image and a method for determining the line direction and line address of the input image, and FIG.
The figure is an explanatory diagram of a method for cutting out characters to be recognized from an input image, and FIG. 4 is an explanatory diagram of a character recognition method. 1... Image input section, 2... Image memory, 3...
Line direction determination section, 4...Character cutting section, 5...Recognition section, 6...Dictionary, 7...Editing section, 8...Text memory.

Claims (1)

[Claims] 1. An image input section that inputs an image containing characters to be recognized, and an image input section that scans the image input in the vertical and horizontal directions to obtain a histogram of pixels forming a character section, and calculates the value of this histogram. A line direction determination unit that determines vertical writing or horizontal writing by comparing the average value of character spacing length defined by the number of consecutive scanning lines with a character spacing of N pixels or less in the vertical direction and horizontal direction, and an input image. A character recognition system comprising: a character extraction unit that extracts a recognition target character from a text; a recognition unit that extracts a recognition candidate character by comparing the recognition target character with a dictionary; and an editing unit that edits the recognition candidate character. Device.