JPS6252337B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention provides a document skew correction device, particularly an optical character reader (OCR) that reads characters on a paper, for example, in a free-format document skew correction device in which no prior information such as character string direction is given. The present invention relates to a document tilt correction device that can automatically correct the tilt of a document screen.

When reading characters using OCR etc., it is first necessary to correct the tilt of the screen. Conventionally,
There are two methods for correcting the paper inclination, such as printing guide marks on the paper in advance to indicate the direction of the characters and using those guide marks as a reference, and methods that detect the paper frame and use it as a reference. Ta. However, in the former case, guide marks must be printed on the paper in advance, and in the latter case, the skew cannot be corrected if the copy is skewed when cut out or copied. It was hot.

In order to solve these drawbacks, the present invention eliminates the need to print guide marks for skew correction on paper that indicate the direction of character strings, and eliminates the need to print guide marks on paper for tilt correction that indicate the direction of character strings. The purpose is to make it possible to correct the inclination even if it does not match the column direction, and will be described in detail below with reference to the drawings.

FIG. 1 is a diagram explaining the principle of tilt correction.
In FIG. 1, 1 is an example of document data to be subjected to skew correction, 2 is a heading, and 3 is one character string in the main text. This document data is raster scanned from a plurality of directions, the electrical signals obtained according to the density are quantized, and the values are accumulated.
When quantized into binary values of 1 and 0, the cumulative value is equal to the length of the black portion on the scanning line. Figure 1 4, 5
are histograms of cumulative density values obtained by scanning in directions 6 and 7, respectively. As is clear from histograms 4 and 5, when the direction in which the characters are lined up matches the scanning direction, a peak 8 clearly appears on the histogram, and this property is utilized to correct the inclination. Specifically, the absolute value of the difference in cumulative density values between adjacent scanning lines is taken,
The sum of the absolute values of cumulative density differences is used, which is the sum of all scans or a portion of scans. Reference numerals 9 and 10 in FIG. 1 indicate the absolute difference values of cumulative density values. Raster scanning is performed on the document screen from a plurality of directions, and the sum of the absolute values of cumulative density value differences is determined for each, and the inclination is corrected so that the value becomes the maximum. When using values for all scans, Perform tilt correction so that where f(l _i )
is the cumulative density value in the i-th scan, and n is the number of scanning lines.

FIG. 2 shows the configuration of an embodiment of the present invention. The quantized density value of the pixel obtained by raster scanning is added to the register 11, and when one raster scan is completed, the difference absolute value circuit 12 is used to add the density cumulative value of the previous scan to the register 11. The absolute value of the difference from the held buffer memory 13 is taken, and the value is added to the register 14. At the same time, the contents of the register 11 are written to the buffer memory 13, and the register 11 is cleared. When raster scanning of the entire screen is completed, the contents of the register 14 are stored in the memory 15 and the register 14 is cleared. An angle correction device 16 looks at the contents of the memory 15 and corrects the angle by rotating the actual screen so that the sum of the absolute values of the cumulative density value differences is maximized.

It is clear that the present invention can be applied not only to the case where input paper is input by changing its angle, but also to the case where quantized data is already stored in an image memory or the like. It is also clear that the same function can be achieved by directly inputting the analog signal to an integrator or the like without quantizing the raster scanning signal.

As explained above, according to the present invention, when reading characters using OCR, even if the paper used does not have guide marks for tilt correction, or even when copying, etc. Even if the copy is tilted, the tilt of the screen can be corrected. Furthermore, although the present invention is largely effective for printed documents, it is also effective for handwritten documents and for screens that include tables, graphs, etc. in addition to text.

[Brief explanation of the drawing]

Figure 1 is a diagram explaining the principle of the device of the present invention, Figure 2
The figure shows the configuration of an embodiment of the present invention. 1...Example of document data to be skewed, 2
... Heading part, 3 ... One character string in the text, 4, 5 ... Histogram of cumulative density values obtained by scanning in directions 6 and 7, 8 ... Existence of character string appearing on the histogram Peaks showing 9, 10
. . . Density cumulative value difference absolute value, 11 . . . A register for adding the quantized density value obtained by raster scanning, 12 . . . Difference between the density cumulative value obtained by raster scanning and the density cumulative value of one scan before. Difference absolute value circuit for calculating the absolute value, 13...Buffer memory for holding the density cumulative value for one scan obtained by raster scanning, 14...Register for adding the difference absolute value of the density cumulative value, 15... A memory for storing the sum of absolute values of cumulative density differences, 16...An angle correction device for rotating the screen and correcting the angle.

Claims (1)

[Claims] 1. In a screen reading device that performs raster scanning on a two-dimensional screen, there is a density accumulating unit that accumulates the value of an electric signal according to the density corresponding to the presence of the screen for each raster scanning line, and A difference absolute value extraction unit that extracts the absolute value of the difference between the accumulated values in the line, a difference absolute value sum extraction unit that extracts the sum of the difference absolute values, and corrects the tilt of the input screen depending on the size of the sum. A document skew correction device comprising: an angle correction section.