JPH09218948A - Method for calculating center of image tending to concentric circle - Google Patents

Abstract

PROBLEM TO BE SOLVED: To accurately find the center of graphic by drawing plural circles with different centers on a binarized object graphic and defining the center of a circle having the maximum or minimum average length overlapping these circles with the binarized graphic as the center of arcuate or radial graphic. SOLUTION: Plural circles with different centers O and P are drawn on the binarized object graphic such as a mesh 17, for example, the stream of picture elements within a range, which crosses one circle, composed of picture elements having continuous density on one side of the binarized graphic, namely, the part overlapping an arcuate object 17a with circles is defined as a segment, and a length L of the overlapped mesh 17 is defined as a segment length. In this case, a value dividing the total of lengths of segments with the number of these segments, namely, a value dividing the total of segment lengths of plural circles with the total number of segments in the case of a concentric circle composed of plural concentric circles is defined as the average length of segments. Then, the center of a circle having the maximum value of this average length of segments is defined as the center of the object graphic.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a concentric circle tendency image center calculation method for calculating the center of a figure that becomes a concentric circle or a figure that spreads radially from one point.

[0002]

2. Description of the Related Art Spiral webs, radial cracks in glass when subjected to point impact, ripples in water, concentric circles or spiral pattern images of fluid vortices and snails are obtained by image processing. Based on the center coordinates, it is necessary to identify the radial direction and the circumferential direction, and to analyze the change state of the density distribution in each direction and binarize it to determine the pitch, line width, line density, ratio of ground to line, etc. May be. In such a case, conventionally, the center position is determined within the image area by visually observing the monitor screen.

[0003]

However, such a graphic often has no center in the strict sense,
In such a state, since it is a work to determine the position which is considered to be the center, there are large individual differences and it takes time. Therefore, it has been desired to automatically determine the center by image analysis.

The present invention has been made in view of the above problems, and an object of the present invention is to provide a method for obtaining the center of a pseudo circular shape or radial figure.

[0005]

In order to achieve the above object, according to the invention of claim 1, a plurality of circles having different centers are drawn on a binarized target figure, and one binarized density is continuous. The pixel row of the range which consists of the pixels and intersects with one circle is set as a segment, and the average value of the segment is calculated by dividing the total value of the lengths of the segments of this circle by the number of the segments,
The center of the circle having the maximum or minimum average length of this segment is the center of the target graphic.

When one density of a binarized figure, for example, a white area has an arc shape, a circle having the same center as the center of the arc body and having the same radius overlaps the arc body.
If the centers are offset, they do not overlap. Many of these arc-shaped bodies are not accurate arcs but have been deformed. So shift the center,
A plurality of circles having the same radius are drawn, and an area of an arcuate body overlapping with one circle is set as a segment, and the average length of this segment is obtained for each circle. The center of the circle having the largest average length of the segment is defined as the center of the arcuate body, assuming that the circle having a large number of overlaps is close to the circle formed by the arcuate body.

On the other hand, in the case where the above-mentioned white area is radial, if the area where the circle drawn at the same center as the center of this radial body overlaps the area of the radial body is taken as a segment, this segment The average length is the smallest compared to the case of circles of the same radius drawn at other center positions. This is because the other central circle diagonally overlaps the radial body, and thus the length of the segment becomes long. Thus, if the figure is determined as a group of arcuate bodies or a group of radial bodies, the center of the figure can be estimated. The radial pattern is the same when the space between the white regions forming the circular arc is large in the radial direction and the probability of intersecting the drawn circular arc is small.

According to the second aspect of the present invention, a plurality of concentric circle groups having different centers are drawn on the binarized target figure, and the binarized pixel group consists of continuous pixels of one density and intersects with one circle. Determine the average length of a segment by dividing the total length of the segments of the circles that make up the concentric circles by the number of segments, with the pixel row in the range being the segment as the maximum value or the minimum value. The center of the concentric circle is the center of the target figure.

A circle drawn to find the center of a figure of an arcuate body or a radial body draws a plurality of concentric circles with respect to one center position, totals the segments of the plurality of circles, and sums the segments. Calculate the average length of the segment divided by the number of, and in the case of an arcuate figure, the one with the largest average length is the center of the figure, and in the case of a radial figure, the figure with the smallest average length is the figure As the center of. By detecting the overlap with respect to the concentric circles in this manner, the accuracy of center estimation is improved. Even if the radial intervals of the arc-shaped white areas are somewhat large, drawing a large number of concentric circles increases the probability of overlapping with any of the arc-shaped white areas.

According to the third aspect of the present invention, when the average length of the circle or concentric circle segment having the center adjacent to the obtained center is close to the obtained center circle or the average length of the concentric segment. Calculates the center of gravity obtained by weighting the calculated center and the center position of the circle having the average length of the approaching segments by the average length of each segment or its reciprocal, and sets this center of gravity as a new figure center.

The shapes of arcuate bodies and radial bodies are irregular, and in many cases cannot be represented by one center. Further, since the center positions of the drawn circles are discrete (discontinuous), the actual center and the drawn center generally do not match.
For this reason, it is often the case that a plurality of average lengths of segments having close values are obtained. In such a case, the center of gravity between the obtained center and the center position of the circle having the average length of the approaching segments is obtained.At that time, in the case of an arcuate body, the average length of the segment is weighted, and in the case of a radial body, The center of gravity is obtained by weighting the reciprocal of the average length, and the center of gravity is set as the new center of the figure.

[0012]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing a configuration example of an image processing apparatus embodying the present invention. The image pickup apparatus 1 has a two-dimensional CCD sensor or the like, and picks up an image of the DUT 14 and outputs an electric signal. The DUT 14 is transferred by a transfer device 16 such as a conveyor. An illumination device 15 is provided in the vicinity of the image pickup device 1, and the object to be measured 14 is transferred.
Illuminate. Illumination is performed from above, from multiple directions, or in a ring shape from the entire circumference.

The A / D converter 2 converts the input data from the image pickup device 1 from analog to digital, and the input buffer 3 temporarily stores the digital data. Bus 4
Transmits a signal, the program memory 5 stores a program that defines the operation of the apparatus, and the CPU 6 controls the entire apparatus according to this program. The image processor 7 performs a gradation process and a binarization process on the input image data. The image analyzer 7a is the most important and essential function of the present invention,
That is, an annular scan address is generated for a binary image, the binary image memory is inspected based on this address, the segment already mentioned is found, and its length is calculated. The binary image memory 8 stores binary image data, and the grayscale image memory 9 stores grayscale image data. The output buffer 10 temporarily stores the data to be output, the D / A converter 11 converts this output data from digital to analog, and the monitor 12 displays this output on the screen. The input device 13 is composed of a keyboard and a mouse and inputs instructions, data and the like.

A method of generating an annular scan address in the image analyzer 7a will be described. Figure 2 shows (X0, Y
0) as the center and addresses S0, S1, ... on a circle of radius r
It is a figure which shows ... Scan addresses are S0, S1, S
2, S3, S4, ..., The distance | Sn + 1-Sn | is set to a unit length. That is, the number of addresses is 2πr / p, where p is the unit length.
Scanning in an annular shape means S1, S2, S
, 3 ... Sequentially occur, and the pixel of the coordinates closest to these is taken out and inspected.

FIG. 3 shows an example of the construction of means for generating such pixel coordinates. The step counter 21 is a counter that generates steps (steps) equal to 2πr / p.
Each step is synonymous with giving the angle θ in FIG. The output of the step counter 21 is the trigonometric function generation RO
It is given to M22 and M23 and decomposed into horizontal and vertical components. Each component is multiplied by the radius r in the multiplier 24 and converted into a concrete distance from the center positions X0 and Y0 of the circle. Further, the adder 25 adds the coordinates of the center position of the circle, X0 and Y0, to obtain the final address on the memory.

In the above description, an example in which a grayscale image is binarized by some method is described as an example in which the grayscale image is scanned in an annular shape. The segment length may be obtained by performing binarization processing on the pixel row obtained as above. In this way, the binarization memory 8 can be substantially omitted, which is more convenient.

A case in which a melon is used as the DUT 14 will be described. In the melon grade evaluation, the fineness of the surface mesh, the fineness of the mesh lines, the uniformity of the mesh, etc. are evaluated. Since such evaluation is performed by the image processing technique, the melon is imaged and the mesh is analyzed. The mesh is approximately circular arc shape (circular arc body) and radial shape (radial body)
If it is configured by, the analysis becomes easy. In particular, if the center of the mesh is used as the origin of the coordinates and handled in association with the coordinates, the analysis becomes easy. Therefore, the center position of the mesh is estimated by the image processing technique.

FIG. 4 represents the surface of the melon imaged with the device of FIG. The melon of the object to be measured 14 is placed on a plate on the transfer device 16 such as a belt conveyor or the like so that the vine part is substantially upward and passes under the imaging device 1. This is illuminated from a plurality of upward directions or from the entire circumference by the ring illumination 15 to obtain images of red, green, and blue, and then binarized to select a color that easily produces a mesh. Shading correction is performed and binarization is performed. Although the image has a vine effect, it is ignored in the actual mesh analysis because the vine effect is small. FIG.
Shows a state in which three concentric circles are drawn around the position O which is considered to be the center of the mesh. In the case of the melon mesh, the circumferential direction is close to a circular arc, but the radial lines are irregular and not radial. Therefore, the center of the mesh is estimated based on the shape of the mesh in the circumferential direction.

FIG. 5 is a diagram showing an overlap between a mesh and a circle centered on a position assumed to be the center of the mesh. The mesh 17 is a circular arc-shaped body 17a in the circumferential direction and a radial body 17b in the radial direction.
However, since the circular arc-shaped body 17a in the circumferential direction is clearer, this will be considered. The degree of overlap between a circle having the same radius and having centers O and P and the arcuate body 17a is shown. A portion where a circle overlaps the arcuate body 17a or the radial body 17b is called a segment, and the length L of the overlapping mesh 17 is a segment length. The circle of the center O largely overlaps the arcuate body 17a in the range of O1 and O2, and the circle of the center P largely overlaps the arcuate body 17a in the range of P1 and P2.

The average length of the segment is defined as an index showing the degree of overlap of one circle with the mesh 17. In the case of one circle, this is the value obtained by dividing the total length of the segments of one circle by the number of that segment. In the case of a concentric circle consisting of multiple concentric circles, It is a value obtained by dividing the total length by the total number of segments of the plurality of circles.

FIGS. 6A and 6B show circles deviated from the center to find the center of the mesh, FIG. 6A shows nine circles having the same radius with the centers deviated, and FIG. 6B shows three circles having different radii. 3 shows three sets in which the centers of concentric circles of For each circle or concentric circles with such a center offset, the length of the segment and the average length of the segment are obtained by image analysis, and when there are many arcuate bodies 17a like the mesh of the melon, the circle with the largest average length of the segment. The center position of is the center of the melon mesh.

In this way, when the center of the mesh or the like is obtained from the center of the circle or concentric circle having the maximum value of the average length of the segments, the average lengths of the segments are about the same and a plurality of adjacent center positions occur. Cases often arise. In such a case, the average length of the segments is weighted to find the center of gravity with the adjacent centers, and this is used as the center of the arcuate body. In the case of a radial body, the center of gravity is weighted by the reciprocal of the average length.

FIG. 7 shows the average lengths of the segments when nine centers shown in FIG. 6A are set and the centers of the arcuate bodies are calculated. The average length of the nine center positions A is 5, which is the maximum, but the average length of the position B immediately above in the Y direction is 4, the average length of the position C on the left side in the X direction is 3, and the three positions are 3 positions. The average lengths of A, B and C are close values. In this case, if the point A is the origin and the distance between A and B and the distance between A and C are 1, the center of gravity can be obtained as follows.

Gx = 3 × 1 / (3 + 5) = 0.375 Gy = 4 × 1 / (4 + 5) = 0.444 The center of gravity is (Gx, Gy), and this center of gravity is the center of the arcuate body. .

FIG. 8 is a diagram showing a case where cracks are radially formed. In this case, the radial body 17 is centered around O.
b has spread. Circle D centered on O is a radial body 17
Although the segment length is short because it is substantially orthogonal to b, the circle E having a center offset from O intersects the radial body 17b at an angle, and thus the segment length is long. Therefore, in the case of a radial body, the center of the circle having the smallest average segment length is the center of the radial body.

[0026]

As is apparent from the above description, according to the present invention, a plurality of circles having different centers are drawn on a binarized target figure, and a portion where the circle overlaps the binarized figure is obtained. , By calculating the average length of overlap for each circle, the center of the circle with the maximum average length is the center of the arc-shaped figure, and the figure with the minimum average length is the center of the radial figure.
It is possible to accurately find the center of a figure of each character. The accuracy of the center position is improved by making the plurality of circles into a plurality of concentric circle groups. If the average value length of the overlap is close to each other with adjacent centers, the center of gravity weighted by the average length of the overlap is the center for the arc-shaped figure,
For a radial figure, centering the center of gravity weighted by the reciprocal of the average length of overlap improves the accuracy of the center position.

[Brief description of drawings]

FIG. 1 is a block diagram of an image processing apparatus that realizes an embodiment of the present invention.

FIG. 2 is a diagram showing addresses S0, S1, ... On a circle centered at X0, Y0 and having a radius r.

FIG. 3 is a diagram showing a configuration example of means for generating pixel coordinates of addresses S0, S1, ... On a circle.

FIG. 4 is a view showing a mesh on the surface of melon.

FIG. 5 is a diagram showing an overlap between a melon mesh and a circle assuming the center of the mesh.

6A and 6B show circles drawn by shifting the center for determining the center of the mesh, wherein FIG. 6A shows nine circles with offset centers, and FIG. 6B shows three sets of concentric circles with offset centers.

FIG. 7 is an explanatory diagram for obtaining the center of gravity when the average lengths of the overlaps of circles having adjacent centers in arcuate figures approach each other.

FIG. 8 is a diagram illustrating a relationship between a radioactive figure and its center.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Imaging device 7 Image processor 7a Image analyzer 14 Object to be measured 17 Mesh 17a Arcuate body 17b Radial body 21 Step counter 22,23 Trigonometric function generation ROM 24 Multiplier 25 Adder

Claims (3)

[Claims] 1. A plurality of circles having different centers are drawn on a binarized target graphic, and a pixel row in a range which is made up of consecutive binarized pixels of one density and intersects with one circle is formed. As a segment, calculate the average value of the segments by dividing the total length of the segments of this circle by the number of the segments, and set the center of the circle where the average length of this segment is the maximum value or the minimum value as the center of the target figure. A concentric circle tendency image center calculation method, characterized by: 2. A pixel row in a range in which a plurality of concentric circle groups with different centers are drawn on a binarized target graphic and which is binarized and consists of continuous pixels of one density and which intersects with one circle. Is the segment and the total length of the segments of the circles that make up the concentric circles is divided by the number of segments to obtain the average length of the segment, and the center of the concentric circle where the average length of this segment is the maximum or minimum value is determined. A concentric circle tendency image center calculation method characterized in that the center of a target graphic is used. 3. If the average length of a circle or a concentric segment having a center adjacent to the obtained center is close to the obtained center circle or an average length of concentric segments, the obtained center is determined. And a center position of a circle having an average length of the approaching segments are weighted by the average length of each segment or its reciprocal, and the center of gravity is determined as a new figure center. Alternatively, the concentric circle tendency image center calculation method described in 2.