KR970008521B1 - Method of detecting horizontal/vertical line of graphics - Google Patents

Abstract

Disclosed is a method to detect horizontal/vertical line in a graphic image. By the method, 4 X 4 mask is duplicated on the left upper side of input image. The number of black pixels in the mask is acquired and compared to a reference("3"). The mask number is increased(boxnum=bixnum+1) to move the mask if the number of black pixels is over the reference. The mask number is compared if the number of black pixels is below the reference. The line number is increased(L=L+1) and the line coordinate is stored, if the mask number is over a maximum value. The above steps are repeated so that horizontal lines are detected. The number of vertical pixels in a mask data is compared to a maximum value so that vertical lines are detected. Thereby, a character in a graphic area can be detected and the line detection time can be reduced.

Description

Horizontal / Vertical Line Detection Method of Graphic Image

1 is a block diagram of a horizontal / vertical line detection system of a graphic image according to the present invention.

2 (a) and (b) is an explanatory view of the image irradiation direction according to the present invention.

3 is an explanatory view of a 4x4 mask according to the present invention.

4 is a horizontal and vertical line detection control flow chart of a graphic image according to the present invention.

Figure 5 (a) to (bar) is a diagram illustrating a horizontal line connection according to the present invention.

6 (a) and (b) are explanatory diagrams of compression data generation for a mask according to the present invention.

7 is a line coordinate explanatory diagram for explaining the present invention.

8 is a horizontal line connection control circuit diagram according to the present invention.

* Explanation of symbols for main parts of the drawings

1: Scanner 2,10: Processor

3,4,7,11: Interface 5,8,13: Memory

6,12 program memory 9: host PC

The present invention relates to a straight line detection method in a graphic image. In particular, the data irradiated during horizontal line detection using a 4 × 4 mask is used for vertical line detection to reduce time and minimize distortion of an input image. A horizontal / vertical line detection method of an image.

In general, in order to detect lines in a document image, the transverse black pixels are counted by one pixel in the transverse direction. The line is detected by determining the number of black pixels while moving one pixel in the direction.

However, this pixel is searched one by one to determine the number of black pixels in one line, and this black pixel number is determined by comparing with an arbitrary threshold value. However, since each pixel is searched, it takes a long time and is weak in noise of an input picture. When an arbitrary pixel of one line is searched to a white pixel ("0") by noise, an error that does not identify the line as a line occurs.

In view of the above problems, the present invention uses a 4 × 4 mask to determine a horizontal line, and uses compressed data obtained from the horizontal line search for the vertical line search to reduce the time for detecting a straight line. Invented a horizontal / vertical detection method, which will be described in detail with reference to the accompanying drawings.

FIG. 1 is a schematic diagram of a graphic image recognition system according to the present invention. As shown in FIG. 1, a microcomputer that separates a graphic region from a scanner 1 that receives a document image and a document image input through the scanner 1 is illustrated in FIG. A processor 1 (2), a memory (5) storing input data, a data memory (8) storing a processing result of the processor 1 (2), a program memory (6) of the processor 1 (2), Processor 2 (10) which extracts a line by using the processing result of the processor (2), a memory (13) for storing the processing result of the processor 2 (10) and a program memory (12) of the processor 2 (10) ), A host PC 9 for displaying the processing results of the processors 1,2 (2) and 10 and the output of the scanner 1 on the monitor, and the processors 1,2 (2) and 10). And interfaces (3), (11) and (4) for interfacing with the scanner (1) and the host PC (9). It was composed of interface 7 for interfacing between the processor 1 and 2 (2), 10.

First, the processor 1 (2) receives the document image data through the scanner 1 and stores it in the memory (5), and obtains the coordinates of the graphics area according to the program operation of the program memory (6) of the data memory (8) The data is stored in the processor 2 and input to the processor 2 10 through the interface 7. The processor 2 (10) obtains horizontal and vertical lines according to its program memory (12) in accordance with program operation, stores them in the memory (13), and interfaces the values of the memory (13) using a host PC (9). (11) can be input and displayed on the monitor.

3),(흑화소수<3)를 압축마스크 데이타 생성부 및 수직라인 검출부에 저장하며, 마스크내의 흑화소수가 임의의 값('3')보다 큰 경우(흑화소수

3)에는 마스크의 수에 해당하는 값(boxnmum)을 증가(boxnum=boxnum+1)시킨 후 마스크를 이동하여 다음 좌표에 마스크를 씌우고 마스크내의 흑화소수를 판별한다. 만약, 마스크내의 흑화소수가 임의의 값보다 작은 경우(흑화소수<3)에는 상기 마스크 수에 따른 값(boxnum)을 임계값('12')과 비교하여 큰 경우(boxnum≥12)에는 라인이 검출되었다고 판별하여 라인수(L)를 증가(L=L+1)시키고, 작은 경우(boxnum<12)에는 그 마스크 수에 따른 값(boxnum)을 초기화(boxnum=0)시켜 라인검출이 아님을 판별하며, 상기 라인(L)검출(L=L+1)후 마스크 내부의 흑화소수가 임의의 값보다 큰 연속된 마스크(boxnum=1∼boxnum≥12)중의 맨처음 마스크의 좌상단 좌표[제3도의 좌표(0,0)]와 맨 나중 마스크의 우하단 좌표[제3도의 좌표(3,3)]를 라인 좌표(H←L)에 저장한다.4 is a horizontal and vertical line detection control flow chart of the graphic image according to the present invention. As shown in FIG. 4, the processor 2 10 extracts a line of the graphic image. When a graphics area is detected by the processor 1 (2) and stored in the data memory 8 while being input through the scanner 1 and stored in the memory 5, the processor 2 10 is stored from the data memory 8. After receiving the coordinates of the graphic area, the width W and the height H of the area corresponding to the coordinate values are obtained from the memory 5 and a 4 × 4 mask is placed on the upper left of the input image. Here, as shown in FIG. 3, a mask having a horizontal coordinate (i = 0, 1, 2, 3) and a vertical coordinate (j = 0, 1, 2, 3) of 4x4 is used. After masking, obtain the black pixel number in the mask, compare the black hydrogen with an arbitrary value ('3'), and compare the black pixel number with an arbitrary value (black pixel number).

3), (black pixel number <3) is stored in the compression mask data generation unit and the vertical line detection unit, and the black pixel number in the mask is larger than an arbitrary value ('3') (black pixel number).

In 3), the value (boxnmum) corresponding to the number of masks is increased (boxnum = boxnum + 1), the mask is moved to mask the next coordinate, and the number of black pixels in the mask is determined. If the number of black pixels in the mask is smaller than an arbitrary value (the number of black pixels <3), the line is formed when the value (boxnum) according to the number of masks is larger than the threshold value ('12'). If it is detected that the number of lines L is increased (L = L + 1), and if it is small (boxnum <12), the value (boxnum) corresponding to the number of masks is initialized (boxnum = 0) to determine that the line is not detected. After the line L detection (L = L + 1), the upper left coordinate of the first mask in the consecutive masks (boxnum = 1 to boxnum≥12) having a larger black pixel number than an arbitrary value [third] Coordinates (0,0) in the diagram and the lower right coordinates (coordinates (3,3) in FIG. 3) of the last mask are stored in the line coordinates (H ← L).

Thereafter, the mask is moved until the end of the full-area irradiation is completed, the line is repeatedly detected, and when the full-area irradiation is completed, horizontal connection is performed. 2 (a) and (b) are explanatory diagrams of the image irradiation direction according to the present invention, and FIG. 2 (a) is the image irradiation direction when the horizontal line is detected, and (b) the image irradiation direction when the vertical line is detected. to be. That is, since the mask is a 4x4 mask, image irradiation is performed while moving by 4 pixels.

6 (a) and (b) are explanatory diagrams of compressed data generation by a mask according to the present invention. When the original image is input and a 4x4 mask is applied as shown in FIG. 4 × pixels of belong to the mask MASK1.

If the number of black pixels in the mask MASK1 is obtained and the number of black pixels is greater than or equal to an arbitrary value ('3'), the black pixel value is determined by the coordinate M1 of the compressed image (a / 4, b / 4) of FIG. The value " 1 " is stored, and if the number of black pixels in the mask MASK1 is smaller than the arbitrary value '3', the coordinate M1 value is stored as " 0 ".

Similarly, the value of the moved mask MASK2 is recorded in the coordinate M2 of the compressed image. In this way, the entire compressed image (a / 4, b / 4) can be obtained by detecting the pixel through the entire image (a, b). When the horizontal line is detected, the compressed image value is obtained. In the vertical line detection, the vertical data is critical with the data of the compressed image as shown in (b) of FIG. 6 even though the mask is not moved vertically as shown in (b) of FIG. After determining whether the value is greater than or equal to the value, the vertical line of the compressed image is detected, and the coordinate of the original image is obtained by multiplying this coordinate value by 4 times to detect the vertical line of the original image.

When the horizontal line is detected, the horizontal line connecting process shows one line as coordinates (x0, y0) and (x1, y1) as shown in FIG. Represented using a two-dimensional array for. That is, the subscript [i] in the line L [i] [j] (i = 1 to n, j = 0,1,2,3) represents an arbitrary line, and the subscript [j = 0,1,2,3 ] Represents the coordinates x0, y0, x1, y1 of the line L [i]. First, the coordinates of an arbitrary line L [i] and its adjacent line L [i ± 1] are compared to determine whether to be connected.

N이 될 때까지 L(i=1) 라인과 Lk를 비교 판별하며 그리고, i=i+1로 증가시켜 i≤N가 될 때까지 상기 과정을 반복 수행하여 처리한다.8 is a horizontal connection control flow diagram according to the present invention. First, the line variables [i] and [k] are set (i, k = 1 to N), and then i = 1 and k = 1, and the difference is' 1 'and [| ik | = 1] are compared and the proximity of lines Li and Lk is determined according to the variables [i] and [k]. If | ik | = 1, the proximity of Li and Lk is determined. It is determined whether the horizontal line is connected by comparing the coordinates of, and in the case of (a) to (bar) of FIG. 5, after adjusting the line coordinates, the coordinates of the line [Lk] are deleted and the line variable [k] is deleted. ] Flag is removed with "0". If not adjacent line [| ik | = 1], increase the value of k when two lines Li and Lk are not the same as (a) to (bar) in FIG. 5 and after coordinate adjustment (k = k + 1). Determine whether the flag of k = k + 1 is "0" and k when the flag [k] is not equal to 0.

The L (i = 1) line and Lk are compared and discriminated until N is reached, and the process is performed by repeating the above process until i = N by increasing i = i + 1.

For example, in case of

L [i] [0] <L [k] [0]. … … … … … … … … … … … … … … Formula (1)

L [i] [3] = L [k] [1] -1... … … … … … … … … … … … … Formula (2)

L [i] [2] <L [k] [2]. … … … … … … … … … … … … … … Formula (3)

Since the above equations (1), (2) and (3) are satisfied, the line coordinate values are adjusted as follows.

L [i] [0] = L [k] [0]. … … … … … … … … … … … … … … Formula (4)

L [i] [1] = L [k] [1]. … … … … … … … … … … … … … … Equation (5)

L [i] [2] = L [k] [2]. … … … … … … … … … … … … … … Formula (6)

L [i] [3] = L [k] [3]. … … … … … … … … … … … … … … Formula (7)

After adjusting as described above, the lines 1i and 1k are identified as one line as shown in the right side of FIG. In the case of (c) to (e) in FIG. 5, the adjustment is discriminated similarly to the right side of (c) to (e) even when the consecutive pixels are 8 pixels or less.

As described above, the present invention detects a horizontal line using a 4 × 4 mask in a graphic region (especially a table region) of a document image, determines and adjusts adjacent lines of the horizontal line, and generates compressed image data. Since the vertical line is detected using the compressed image data, it is possible to recognize the characters in the graphic area and to shorten the linear detection time.

Claims (2)

Obtain the width (W) and height (H) of the graphics area, cover the 4x4 mask with the top left corner of the input image, find the number of black pixels in the mask, and determine whether the black hydrogen is greater than or equal to a random value ("3"). Compressed mask data is stored and the mask value is increased (boxnum = boxnum + 1) when the number of black pixels in the mask is greater than or equal to the random value, and the mask is moved to the threshold when the number of black pixels is less than the random value. Determine and exceed threshold (boxnum

12), the number of lines is increased (L = L + 1), the line coordinates are stored, when the value is less than the threshold value, initialization is performed (boxnum = 0), and the entire area is examined while moving the mask. After detecting the lines, the horizontal lines are connected to detect the horizontal lines, and the horizontal / vertical lines of the graphic image which detects the vertical lines by determining the vertical pixel number of the vertical lines of the compressed mask data as a threshold value ('12'). system. The method of claim 1, wherein the x-coordinate value difference between two horizontal lines of the same size is equal to or less than a random value, and the two horizontal lines of the same size are equal to or less than an arbitrary pixel (8 pixels). And a horizontal / vertical line detection method for connecting the adjacent horizontal lines when the y-coordinate difference is one.

Images