JPH05266250A - Character string detector - Google Patents

Abstract

PURPOSE:To provide a character string detector capable of detecting required character strings at a high speed without performing excess recognition processing such as deleting objects not to be recognized. CONSTITUTION:This detector is provided with as pixel selection means 1 selecting either of white and black picture elements among a binary pictures, means 2 scanning image data in a transverse direction and determining the number of picture elements in the vertical direction of the picture element selected by the means 1, means 3 scanning the image data in a longitudinal direction and determining the number of picture elements in the horizontal direction of the picture element selected by the means 1, coordinate detecting means 4 detecting the four coordinate points where the difference of the picture element number becomes maximal, a verifying means 5 comparing the coordinate value with the threshold value and verifying whether it is the size of the character string to be processed.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a character string detecting device,
In particular, the present invention relates to a character string detection device that searches a character string in a document and detects the character string when the character string is read in the document to recognize only a specific character string.

[0002]

2. Description of the Related Art In order to detect a specific character string from binary image data, a conventional character string detecting device must add a specific mark to the beginning of the character string and recognize and detect the mark. , There is a method of detecting a character string (hereinafter, a mark detection method).

Further, the character strings are classified based on the information that the character strings in the image mean, and the character strings are recognized by character recognition.
There is a method of understanding and extracting a processing target character string (hereinafter referred to as a character string understanding method). This is a method of recognizing all the characters in an image and extracting a character string recognized as a number when only the target character string is composed of numbers, for example.

[0004]

However, in the conventional techniques, in order to detect a specific character string from a binary image, a character is detected from the image in both the mark detection method and the character string understanding method. It is necessary to extract a character rectangle by cutting out in units of one character and perform character recognition for all character rectangles. For this reason, since character cutting and recognition processing are performed even for characters that originally do not need to be recognized, there is a problem that a redundant processing amount exists and processing time increases.

On the other hand, if there are ruled lines, decorations, etc. in the image, it is not necessary to recognize them, so a process for removing the ruled lines, etc. is required, which causes an increase in processing time.

The present invention has been made in view of the above points,
An object of the present invention is to provide a character string detection device capable of detecting a necessary character string at high speed without performing an extra recognition process and erasing an unrecognized object.

[0007]

According to the present invention, processing is performed from image data of a binary image which is expressed by distinguishing a character string to be processed in advance inputted from an image input device and a character string not to be processed. In a character string detecting device for detecting a target character string, a pixel selecting means for selecting one of a pixel and a black pixel in a binary image, and a pixel selecting means for horizontally scanning image data. Means for obtaining the number of pixels in the vertical direction of the pixels selected by, and means for scanning the image data in the vertical direction to obtain the number of pixels in the horizontal direction of the pixels selected by the pixel selection means; The size of the character string to be processed by comparing the coordinate value of the rectangle obtained by the coordinate detecting unit with the predetermined threshold value, and the coordinate detecting unit that detects four coordinate positions where the difference in the number of position pixels is maximum. It And a test means to test whether or not there.

Further, according to the present invention, when the black pixel is selected by the pixel selecting means, the coordinate detecting means determines the number of black pixels in the vertical direction at the first arbitrary position and the direct number of the first arbitrary position. The number of black pixels in the vertical direction at the first arbitrary position and the second arbitrary position where the difference in the number of black pixels in the left vertical direction is the maximum, and the number of black pixels in the vertical direction immediately to the right of the second arbitrary position The second arbitrary position having the maximum difference from the black pixel number, the horizontal black pixel number at the third arbitrary position, and the horizontal black pixel number immediately above the third arbitrary position. The third arbitrary position where the difference is maximum and the horizontal black pixel number at the fourth arbitrary position and the horizontal black pixel immediately below the fourth arbitrary position where the difference is maximum. The respective coordinates of the rectangle are obtained from the respective arbitrary positions of 4.

[0009]

According to the present invention, a character string to be processed, or
The unprocessed character string is printed in white characters in advance, and this is scanned in the horizontal and vertical directions with respect to the image image data input by a scanner, etc., and the number of black pixels or the number of white pixels in the vertical direction, respectively. The number of pixels of one of the two and the number of pixels of the black pixels or the number of white pixels in the horizontal direction are obtained, and a rectangular area surrounded by arbitrary four positions is detected as a processing target area. Character cutting and character recognition are performed only for the processing target area. As a result, it is not necessary to cut out characters in a portion other than the processing target area or perform character recognition, so that processing time is not wasted.

[0010]

Embodiments of the present invention will be described below with reference to the drawings. In this embodiment, it is assumed that the character string to be processed is outlined.

FIG. 1 is a block diagram showing the system configuration of an embodiment of the present invention. The character string detection device of the present invention selects either a white pixel or a black pixel, and scans the image selection data 1 (the black pixel is selected in the present embodiment) and the image image data in the horizontal direction. A horizontal direction histogram generation unit 2, a vertical direction histogram generation unit 3 that scans image image data in the vertical direction, and a maximum difference detection unit 4 that detects a maximum difference in the number of black pixels in the vertical direction and the horizontal direction in any one. , By the rectangle obtained by the maximum difference detection unit 4,
It is composed of an area inspection unit 5 which detects a rectangle to be processed.

FIG. 2 is a diagram for explaining the operation of one embodiment of the present invention. The operation of the character string detection device of the present invention will be described with reference to FIG.

First, before the operation of the character string detection device, the input image image data is such that the character string to be processed is printed in white characters in advance, and when the character string is input by a scanner or the like, the pixel The selection unit 1 selects a black pixel to be a processing target.

The horizontal histogram generation unit 2 scans the binary data input by an image input device such as a scanner in the horizontal direction (direction (X) in FIG. 2). Along with this scanning, the number of black pixels existing in each line is counted in the vertical direction. Shown in FIG. 2a is the counted X-direction histogram.

Next, the vertical histogram generation unit 3
Scans in the vertical direction (Y direction in FIG. 2) and horizontally counts the number of black pixels existing in that line for each line, and FIG. 2B shows the counted Y direction histogram. Is.

When the number of black pixels at any adjacent position i, j (where i is the left side) in the X-direction histogram a is Xblock (i), Xblock (j), the maximum difference detection unit 4 defines Xblock ( The position where j) −Xblock (i) is maximum is stored in the left end coordinate X1 (FIG. 2), and the position where Xblock (i) −Xblock (j) is maximum is stored in the right end coordinate X2 (FIG. 2).

In the difference graph of the X-direction histogram shown in FIG. 2C, e is Xblock (j) -Xblock.
k (i) is the maximum position, and f is Xblock
(I) − This is the position where Xblock (j) is maximum.

Similarly to the above, the maximum difference detection unit 4 also determines the number of black pixels at arbitrary vertically adjacent positions m and n (where m is the upper side) in the Y-direction histogram b as Yblock.
(M), Yblock (n), the position where Yblock (n) −Yblock (m) is the maximum is stored in the upper end coordinate Y1 (FIG. 2), and Yblock (m) −Yblock (n) is the maximum. Is stored in the bottom coordinate Y2 (FIG. 2).

In the graph of the difference in the Y-direction histogram shown in FIG. 2d, g is Yblock (n) -Yb.
lock (m) is the maximum position, and h is the maximum Yblock (m) −Yblock (n) position.

As a result, the upper left coordinate (X1, Y1)
Then, a rectangular region p surrounded by the lower right coordinates (X2, Y2) can be obtained.

Next, the area inspection unit 5 determines the rectangular area p.
The horizontal (X2-X1) and vertical (Y1-Y2) sizes are calculated and compared with predetermined thresholds (Sx, Sy) of the horizontal size and vertical size that are appropriate for the character area, and (X2-X1)> If Sx (Y1−Y2)> Sy and the horizontal / vertical size is larger than the threshold value, it is determined that the character string is to be processed, and the detection process ends.

On the other hand, if the horizontal / vertical size is (X2-X1) <Sx (Y1-Y2) <Sy, the process returns to the process of the maximum difference detection unit 4, and the newly obtained maximum is removed except for the region obtained first. The rectangular area is recalculated by the difference.

FIG. 3 shows a case of re-obtaining a rectangular area according to an embodiment of the present invention, and shows a case where ruled lines are present.

In the first processing, the ruled line r is set in the processing target range z.
Therefore, the rectangular area t surrounded by the upper-left coordinates (Xa, Ya) and the lower-right coordinates (Xb, Yb) from the maximum difference of the Y-direction histogram is determined by the area test unit 5 to have an appropriate lateral size as a character area. Vertical size threshold (Sx,
When compared with Sy), in particular, the horizontal size Xb-Xa is larger than the character size, and therefore the threshold size or more, but the vertical size Yb-Ya is smaller than the character size. Therefore, the area inspection unit 5 determines that the coordinates of the upper end and the lower end are incorrect. As a result, the position having the maximum difference is searched for excluding Ya to Yb, and the upper end coordinate Yc and the lower end coordinate Yd are obtained.
As a result, the upper left coordinates (Xa, Yc) and the lower right coordinates (X
b, Yd), and the rectangular area p is obtained. This rectangular area p has a vertical / horizontal size of (Xb-Xa)> Sx (Yc-Yd)> Sy, and if the horizontal / vertical size is larger than the threshold value, it is determined as a character string to be processed and detected. The process ends.

As a result, the rectangular area p to be processed becomes the target of recognition processing and the like, and the ruled line r has a vertical size smaller than the threshold value. Is not determined.

Although the present embodiment is premised on that the character string to be processed is printed with blank characters, the present invention is not limited to this example, and conversely, characters not treated The columns may be outlined characters. In this case, when the white pixel or the black pixel is selected, the white pixel is selected in advance so that the white pixel is used instead of the black pixel of the above-described embodiment to perform the same processing as that of the above-described embodiment. Therefore, the same effect can be obtained.

In the prior art, it was necessary to perform character segmentation and recognition processing even for a character that does not need to be recognized in order to detect a specific character string. According to the character string detection device, it is possible to detect the character string to be processed in the black pixel histogram by simply printing the character string to be processed in white, and the time for character string detection is greatly reduced. it can.

On the other hand, when there are ruled lines, decorations, etc. in the image, the ruled lines are not removed by comparing with the character string area size determined by the character size.

[0029]

As described above, according to the present invention, it is not necessary to cut out a character not to be processed and perform recognition processing,
Processing time is reduced.

Further, the process of removing unnecessary things such as ruled lines becomes unnecessary, and a desired character string can be detected at high speed from the image data.

[Brief description of drawings]

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

FIG. 2 is a diagram for explaining the operation of the embodiment of the present invention.

FIG. 3 is a diagram showing a case of re-obtaining a rectangular area according to an embodiment of the present invention and a case where ruled lines exist.

[Explanation of symbols]

1 Pixel selection unit 2 Horizontal direction histogram generation unit 3 Vertical direction histogram generation unit 4 Maximum difference detection unit 5 Area test unit a Counted X direction histogram b Counted Y direction histogram c Graph of X direction histogram difference d Y direction histogram Difference graph e Xblock (j) −Position where Xblock (i) is maximum f Xblock (i) −Position where Xblock (j) is maximum i Any adjacent position in the X direction histogram a j In the X direction histogram a Arbitrary adjacent positions m Arbitrary upper and lower adjacent positions n Arbitrary upper and lower adjacent positions p Rectangular area r Ruled line t Rectangular area

Claims (2)

[Claims] 1. A character for detecting a character string to be processed from image data of a binary image which is expressed by distinguishing a character string to be processed in advance inputted from an image input device and a character string not to be processed. In a column detection device, a pixel selection means for selecting either a white pixel or a black pixel in a binary image, and a pixel selected by the pixel selection means by horizontally scanning the image data. Means for obtaining the number of pixels in the vertical direction, a means for scanning the image data in the vertical direction, and obtaining the number of pixels in the horizontal direction of the pixels selected by the pixel selecting means, Coordinate detecting means for detecting four coordinate positions where the difference in the number of pixels is maximum, and the size of the character string to be processed by comparing the rectangular coordinate values obtained by the coordinate detecting means with a predetermined threshold value. Is String detecting apparatus characterized by having a test means for test. 2. When a black pixel is selected by the pixel selecting means, the coordinate detecting means determines the number of black pixels in the vertical direction at a first arbitrary position and the left of the first arbitrary position. Of the number of black pixels in the vertical direction at the first arbitrary position and the second arbitrary position where the difference in the number of black pixels in the vertical direction becomes maximum,
At the second arbitrary position where the difference between the right arbitrary vertical position and the number of black pixels in the vertical direction becomes maximum, and at the third arbitrary position, the number of horizontal black pixels and the third arbitrary position. Of the number of black pixels in the horizontal direction immediately above the position, and the number of black pixels in the horizontal direction at the fourth arbitrary position and the fourth arbitrary position. 2. The character string detection device according to claim 1, wherein each coordinate of the rectangle is obtained from each position of the fourth arbitrary position where the difference with the black pixel in the horizontal direction immediately below is the maximum.