JPH01142874A - Original area detector - Google Patents

Abstract

PURPOSE:To attain original area detection prevented from being influenced by noise or the like by detecting the representative points of respective sides by the scanning of the whole area of an original reading face and prescribed processing capable of neglecting fine white blocks corresponding to noise or the like and detecting the coordinates of four corner points of the original from respective coordinates of the representative points. CONSTITUTION:A side representative coordinate detecting part 6 scans image data along a determined side detecting line from the outside of the original area of the image data, sets up a picture element first detected out of picture elements having the ground color of the original or the density of color on the outside of the original area at the time of detecting the picture elements as the representative point and detects the coordinates of eight side representative points consisting of each two point of respective four sides. An original corner point coordinate calculating part 7 sets up four intersecting points obtained by crossing two lines out of four straight lines connecting respective representative points from the coordinates of eight side representative points as four corner points of the original and calculates and outputs the coordinates of the corner points. Since always stable corner points of an original can be obtained without being influenced by noise or the like, information such as the size of the original, its arranged position and inclination can be accurately obtained.

Description

DETAILED DESCRIPTION OF THE INVENTION (Industrial Field of Application) The present invention relates to a method for detecting an area of a manual document in a manual image processing apparatus.

(Prior art) In a conventional manual image processing device, in which the area of a manual document is detected, the document reading surface is placed so that the document can be distinguished from the surroundings, and is scanned in the main scanning direction and the sub-scanning direction by an image sensor. Create a peripheral distribution from the image data obtained by quantizing it, determine the coordinates of the edge detection line from the peripheral distribution, scan the edge detection line in the image data from outside the document area, and first calculate the coordinates corresponding to the ground color on the document. By detecting the point where the white bit is detected as one of the points representing one side of the document, the coordinates of the side representative points (two points on each side, eight points in total) are detected, and the corner of the document is determined from the coordinates of the side representative points. Point coordinates were calculated and used to represent the document area.

(Problem to be Solved by the Invention) However, in the above conventional method, the edge detection line is scanned from outside the document area and the point where the white bit is first detected is detected as one of the points representing one side of the document. Because
If even one bit of white dot exists on any side detection line outside the document area due to noise etc., a point far away from the actual side may be detected as the 9th side representative point, and as a result, the actual document corner point may be detected as the 9th side representative point. There is a problem in that points that are significantly different from the above are detected as corner points. Here, in FIG. 3 showing the problems of the prior art, A is a blank document, B is a white block such as noise, TI, T2. Ll, L2゜Bl,
B2. R1 and R2 are the detected side representative points, 'tl +Pt
r+Pbl 1Pbr indicates the detected upper left corner point, upper right corner point, lower left corner point, and lower right corner point. As you can see from the same figure,
Due to the noise of white block B, the detected side representative point B,
is far away from the actual lower side, and as a result, the detected lower left corner point Pbn and lower right corner point Pbr are located at positions significantly different from the actual corner points.

The present invention is intended to solve these problems, and it is an object of the present invention to provide a stable document area detection device that is not affected by noise or the like through simple processing.

(Means for Solving the Problems) In order to solve the above-mentioned problems, the present invention reads the document reading surface on which the document is placed in a state where it can be distinguished from the surroundings, and stores image data obtained by quantizing the document. , a document area detection device that detects a document area based on this image data includes a peripheral distribution creation section that creates a peripheral distribution based on the image data, and two points each representing each side based on the peripheral distribution. an edge detection line coordinate determination unit that determines the coordinates of the edge detection line for detecting the edge detection line; and an edge detection line coordinate determination unit that scans the edge detection line from outside the document area and first successively determines the ground color of a predetermined number of documents or the density of the color outside the document area. When a pixel with is detected, the first pixel detected among those pixels is 2
A side representative point coordinate detection unit that detects the coordinates of a total of 8 side representative points, 2 points on each side, and 2 of the 4 straight lines connecting each representative point from the coordinates of the 8 side representative points. Intersection of books 4
The present invention is characterized in that the points are four corner points of the document, and that the document corner point coordinate calculating section calculates the coordinates of the corner points.

(Operation) According to the present invention having the above configuration, the peripheral distribution creation section creates peripheral distributions in each of the horizontal and vertical directions. The side detection line coordinate determination unit determines the coordinates of side detection lines for detecting two points representing each side based on the created peripheral distribution. Then, the side representative coordinate detection unit scans the image data from outside the document area along the determined side detection line, and first continuously obtains a predetermined number of consecutive document ground colors or densities of colors outside the document area. When a pixel is detected, the first pixel detected among those pixels is used as the two side representative points on each side, and the coordinates of the side representative points, two points on each side for a total of eight points, are detected for each side of the document. Detect the coordinates of a total of 8 points (2 representative points on the side). The document corner point coordinate calculation unit uses the coordinates of the 8 side representative points to define the 4 points of intersection where 2 trees intersect among the 4 straight lines connecting each representative point as the 4 corner points of the document, and calculates and outputs the coordinates of these corner points. do.

Therefore, the present invention can solve the above-mentioned problems, and can provide a stable document area detection device that is not affected by noise or the like through simple processing.

(Example) Embodiments of the present invention will be described below based on the drawings.

FIG. 1 is a block diagram showing the configuration of an embodiment of the present invention. In the figure, reference numeral 1 denotes optical input, which is light input from the original reading surface on which the original is placed so as to include the entire surface of the original, which is a quadrilateral. is configured. 2 is a photoelectric conversion unit, which converts the entire area of the document reading surface of optical input 1 into 2
The image is decomposed into 048x3072 images and each pixel is converted into a binary digital signal (hereinafter, this signal is referred to as input image data). 3 is an image memory that stores input image data in a format that can reproduce the X and Y coordinates of each pixel in the entire area of the original reading surface, and has a capacity of 2048 x 3072 bits corresponding to the entire area of the original reading surface. It has the following. 4 is a marginal distribution creation unit, and image memory 3
The human image data is projected onto the X and Y coordinates, that is, in the horizontal and vertical directions in the two-dimensional coordinates of the image memory 3, to obtain the distribution of, for example, the number of black bits, and the marginal distribution S
(x) and S Y (y) are created and stored in a marginal distribution memory (not shown). Reference numeral 5 denotes a side detection line coordinate determination unit which refers to the peripheral distribution memory and determines the coordinates of the side detection line for detecting two points representing each side. Reference numeral 6 denotes a side representative point coordinate detection unit which scans the side detection line indicated by the coordinates of the side detection line determined by the side detection line coordinate determination unit 5 in the image memory 3 and detects a total of 8 side representative points, 2 points on each side. This is to detect. Reference numeral 7 denotes a document corner point coordinate calculation unit, which calculates the side representative point 8 detected by the side representative point coordinate detection unit 6.
From the coordinates of the points, four points of intersection where two trees intersect among the four straight lines connecting each representative point are regarded as four corner points of the document, and their coordinates are calculated and output to the document corner point coordinate output terminal 8. In addition, the marginal distribution S X (x), S obtained by the marginal distribution creation unit 4
Y (!/) is expressed by the following formula.

However, x, y are two-dimensional coordinates in the image memory 3, P (x, y) means a black bit or a white bit, and the black bit (indicating the color of the character part on the document and the color outside the document area) ), P (x, y) = 1, and in the case of a white bit (indicating the background color on the original), P (x, y) = O.

Next, the operation of this embodiment is shown in FIG. 1, and FIG. 2 shows a document reading surface where a blank document A is placed as an example of human image data, and a white block B due to noise exists outside the document area.
Marginal distribution S for the image data of a) and Fig. 2(a)
Figure 2 (b) showing X (x), S Y (y)
, (C).

First, the optical input 1 from the document reading surface on which the original including the entire surface of the quadrilateral original is placed is manually input to the photoelectric converter 2, and the photoelectric converter 2 converts the manually input optical input 1 into input image data. and stored in the image memory 3. The image memory 3 stores the X and Y coordinates of each pixel in the original reading area in a format that allows reproduction of the X and Y coordinates. Then, the marginal distribution creation unit 4 projects the input image data of the image memory 3 along the X and Y axes to create a marginal distribution S
(x).

Create S Y (y) and store it in the marginal distribution memory.

Then, the side detection line coordinate determination unit 5 first determines the threshold values X THL and Y THL in each peripheral distribution as shown in FIG.
).

Determine as shown in (C). This threshold value X THL, Y
THL is expressed by the following formula. ' XTHL = (M-XMIN+N-XMAX)/(M
+N) ---(:1)YTHL = (M-YMI
N+N-YMAX)/(M+N) ---(4) However, XMIN and XMAX are the minimum and maximum values in the marginal distribution S is the maximum value, M,
N is a constant that determines the threshold value. Here, in this embodiment, M=N=1.

Next, in the marginal distribution S X (x), the threshold value X THL
The minimum coordinate x1 and maximum coordinate x2 that are less than
The minimum coordinate Y1 and maximum coordinate Y2 that are less than the threshold value YTHL in Y (y) are detected. Furthermore, the coordinates x1, x2,
Side detection line coordinates X, , X using Y8, Y2 and constant α
s2. Determine Y, , Y, □. The constant α is the coordinate Xs,
, Xs2 are set so that both the upper and lower sides are crossed, and the two side detection lines represented by the coordinates Y, , Yg2 are set so that they both cross the left and right sides.

Therefore, the relationship is as shown in the following equation.

XSI"XI+α...(5) YSI=Yl+α ---(6) xs□=×2−α...(7) Ys2=Y2−α...(8) Here, in this embodiment, α=400.

Note that instead of obtaining the edge detection line from the peripheral distribution of black bits as described above, we create the peripheral distribution of white bits and use the formula (
3) Find the threshold values X THL and Y T from the calculation formula in which the constants M and N in (4) are replaced, and find the minimum coordinate X that exceeds the threshold value X THL in each marginal distribution.

and the maximum coordinate x 2, the minimum coordinate Y1 that exceeds the threshold Y THL
and the maximum coordinate Y2, and from these coordinates, the formula (
Similar side detection line coordinates can also be obtained by determining the side detection line coordinates using 5), (6), (7), and (8).

Then, the side representative point coordinate detection unit 6 first detects the side detection line coordinates X8. ．． Each side detection line indicated by Xs2 is scanned from y=Q to y=3071, and first, a predetermined number of
Here, when 16 consecutive white bits are detected,
The coordinates (X
s', Ytt), (XS2°Yt2). These two points are two points representing the upper side. Furthermore, the side detection lines indicated by the side detection line coordinates Xs,,
X 32. Y bz) is detected. These points are two points representing the lower side. Next, the side detection line coordinates YS,, Y
The side detection lines indicated by S2 are respectively from X=O to x=
Coordinates of the white bit (Xtt, Ys+), (Xtz, YS2) scanned to 2047 and detected using the same method as described above.
Detected. These points are two points representing the left side. Furthermore, the side detection lines indicated by the side detection line coordinates YsI and Ys2 are scanned from x = 2047 to x = O, respectively, and the white bit coordinates (Xr++
Ys+), (X r2.Y S2) are detected. These two points are two points representing the right side. A total of eight side representative points, two representative points for each of the above-mentioned upper, lower, left and right sides, are detected.

If the edge representative points are detected using the method described above, the result shown in Figure 2 (
Even if a white block such as white block B in a) exists outside the document area due to minute noise of 15 bits or less on the side detection line, this can be ignored, allowing more stable side representative point detection. becomes possible.

Then, the document corner point coordinate calculating section 7 in FIG. intersection point 4
The points are defined as four corner points of the document, and the coordinates are calculated using the following calculation procedure.゛First, the coordinates of the representative point on the upper side (Xs+, Ytt), (XS2
゜Yt2) Left side (7) Representative point coordinates (x, , ys,)
, (X12°YS2) to the coordinates of the upper left corner point of the original (X□,
Ytt).

Xtn=((Xs2−Xs+)(Xt2Ys+−Xt+
Ys+)(Xn2Xm+)(Xs2Yt+ Xs+Y
t2))/ ((X12 X1l) (Yt2Yt
+) (XS2 X5I) (Y!12 YSI)
)...(9) Ytu= ((Yt2 Ytt) (Ys2Xt
+ YSIX12) - (YS2 YSI) (Y
t2Xs+ Y t+X52))/ ((YS2-
YSI) (XS2 X5t) - (Yt2 - Ytt
) (X12 Xtt) ) '...(1
0) Below, the coordinates of the representative point of the upper side (Xs+, Ytt).

(X s2. Y t2) and the coordinates of the representative point on the right side (Xr+
, Ys+).

(X r2. Y 82) and the coordinates of the upper right corner point of the original (
Xtr.

Yt, , ) is calculated, and the coordinates of the representative point on the lower side (Xs+, Y
b+), 2 (XS2. Yb2) and the coordinates of the representative point on the left side (
X support 1, Y S I) + (X 12. Y S
2) to the coordinates of the lower left corner point of the original (Xbx).

Yhn), and the coordinates of the representative point on the lower side (Xs+, Yb
+).

(XS2. Yb2) and the coordinates of the representative point on the right side (xr+,
YSI).

(X, 2. Y s2) and the coordinates of the lower right corner point of the original (
xbr.

ybr).

The corner point coordinates of the document calculated in this manner are output to the document corner coordinate output terminal 8, and are used for detecting the inclination of the document as described below.

If the inclination of the upper side is the inclination of the original, and if that inclination is θ, then θ = jan-'((Yt, Ytn) /
(Xtr-Xtt) )...(11) It is given by: It is also possible to use the corner point coordinates and the inclination θ in the process of detecting character strings on a document.

As described above, according to this embodiment, the representative point of each side is detected by scanning the entire area of the document reading surface and a predetermined process that can ignore minute white blocks corresponding to noise, etc., and the representative point of each side is detected. It detects the coordinates of four corner points on the document.
Stable document area detection that is not affected by noise etc. can be achieved with simple processing.

(Effects of the Invention) As described above, according to the present invention, when detecting side representative point coordinates by scanning a side detection line determined from the peripheral distribution of an image including a document, Only when a white bit corresponding to the background color on the original is detected, the point of the white bit detected first among the white bits is detected as the edge representative point. Therefore, noise on the image sensor, original reading surface, etc. Even if a block of white bits occurs outside the document area due to reasons such as this, it can be ignored as long as it is not too large. Therefore, it is possible to detect a document corner point that is strong against noise and is stable at all times, so that it is possible to obtain information such as the size of the document, the arranged position, and the tilt.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the configuration of an embodiment of the present invention, FIG. 2(b) and 2(C) show the marginal distribution S X (X) for the image data of FIG. 2(a). 5Y(y), and FIG. 3 is a diagram illustrating the problems of the prior art. 1... Light input, 2... Photoelectric conversion section, 3... Image memory, 4... Surrounding distribution creation section, 5... Side detection line coordinate determining section, 6... Side representative point coordinates Detection unit, 7... Original corner point coordinate calculation unit, 8... Original corner point coordinate output terminal.

Claims (1)

[Claims] Document area detection that stores image data obtained by reading and quantizing a document reading surface on which a document is placed in a state where it can be distinguished from the surroundings, and detects an area of the document based on the image data. The apparatus includes: a marginal distribution creation unit that creates a marginal distribution based on the image data; and an edge detection line coordinate that determines coordinates of edge detection lines for detecting two points representing each edge based on the marginal distribution. a determining unit; and two side representatives representing one side of the first detected pixel when a predetermined number of pixels are first consecutively detected by scanning the side detection line from outside the document area; a side representative point coordinate detection unit that detects the coordinates of a total of 8 side representative points, 2 points on each side; and 2 of the 4 straight lines connecting each side representative point from the coordinates of the 8 side representative points. Let the four points of intersection be the four corner points of the manuscript,
A document area detection device comprising: a document corner point coordinate calculating section that calculates the coordinates of the corner point.