JPS58190708A - Reference position deciding system in x-y two-dimensional picture - Google Patents

Abstract

PURPOSE:To enhance the precision of determination of the reference position by a method wherein wind regions are set through selection of positions suitable for determination of coordinates in X and Y directions in view of the shape of the picture, and coordinates are determined by making use of binary-coded scanning output signals. CONSTITUTION:A wind region W2 (a,b,c,d) is set at a position suitable for determination of Y coordinates value, while another wind region W1 (e,f,g,h) is set at a position suitable for the determination of the X coordinates value. In the region W2, the picture 7 is scanned in the X-axis direction for each Y coordinates value, and the lengths of segments of this horizontal scanning are measured. The Y coordinates value YSY of the reference position is determined as the Y coordinates value which is the fourth as counted from the Y coordinates value at which a predetermined threshold value is exceeded from the first time. Then, X coordinates values of the rising points in the wind region W1, along the vertical scanning lines at both sides of a Y coordinates value YDX. The mean value of these X coordinate values is determined as the reference position.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention provides orthogonal X-
This relates to a reference position determination method in the X-72 dimension 1ilii4a that defines the coordinates of a reference position in the Y2 dimension.

The eleventh is an explanatory diagram of the imaging screen. In the same, 1#
'1 Screen #1 taken by the iTV camera, #2 is an object image. Screen 1 is represented by mutually orthogonal X coordinates, yi*. The camera performs horizontal scanning f in the X direction and vertical scanning f in the Y direction.

Scan the screen along the straight line whose Y coordinate is Yl.

The waveform of the imaging signal obtained when doing so is as shown in FIG. 2 (0), and the E ratio output is as shown in (2).

Now, in the w&1 diagram, the angle of the object image 2 within the screen 1 can be represented, for example, by a point 8 at the upper left corner of the image 2. In other words, if the X coordinate and : There is. For example, let us assume that rectangular objects of the same shape as shown by 2 of the first decoy are being sequentially conveyed on a conveyor belt. At this time, the ITV camera first takes an image of a good object and stores the imaging result in the memory, and when the next time the transported object is imaged, the imaging result and the imaging result stored in the memory are used. By comparing p1 with p1, it is possible to inspect whether the transported object is a good product with no scratches or not.

When comparing for such tests, *? If there is a discrepancy between the in-plane position when a non-defective product is imaged at 71 and the screen circular image position when the next image of the transported object is taken, a simple comparison may incorrectly determine a non-defective product as a defective product. It can happen. In such a case, if the coordinates of the point S as the reference point of the object image 20 in the first factor can be easily found in the image of the transported object, then these coordinates? Since the image position can be corrected by using Conventionally, the reference position is determined by scanning the iTV camera horizontally and vertically as shown in FIG. , and the Y coordinate Y8 is defined as the coordinate of point S, which is the reference position.

However, this conventional method has the following drawbacks. First, the X coordinate of the starting point S of the object image is determined by detecting the rising position of the imaging signal, so if noise is included in the imaging signal, the position where the noise appears is determined as the X coordinate of the starting point S. It has been published in four editions and has the disadvantage of being sensitive to noise. This will be explained with reference to No. 3-.

Figure 3 shows an example of a photographed fa screen containing noise.
It is a diagram. In the figure, 3 indicates noise and 4 indicates an object image. As is clear from the figure, there is an object @
If Noise 3 is located outside 40 at a position such as O shown in the diagram, Noise 3
Y coordinate Y1 and X coordinate a1 obtained when performing t-scanning
(al indicates the X coordinate of the tip of noise 3) is mistakenly determined to be the coordinate of the starting point of the object image.

The second drawback of the conventional method is that the X coordinate of the starting point of the object image is
Since the coordinates are determined at the found position, the error in the X coordinate becomes large depending on the shape of the object image O. This will be explained with reference to Figure JR5.◎ As seen in 5th all, when the object image 5 has a convoluted pincushion shape in the X direction at its upper end, the X direction scanning line If the X coordinate position of is shifted by just one line from Yl to Yl, 4. The X coordinate position varies significantly from xl to xl. In this rounded object image with such a shape, there is a possibility that the error in the X coordinate of the starting point of -t becomes large. Objects similar to this shape include various objects such as connectors with a large number of protruding bottles, IC (integrated circuit) parts, and feather-like objects.

The drawback that the error in the X coordinate becomes large applies not only to the shape of the object shown in Fig. 5, but also to the circular object image 6 as shown in Fig. N6. Please note that the amount of change jX in the X coordinate is large compared to the amount of change Y.

In general, it is difficult to uniformize the illumination throughout the camera field of view (imaging screen).Due to these circumstances, depending on the location within the object image area, the signal level of oneself and the signal level of the background may differ. Alternatively, there are places where the signal level of the noise is about the same, and in such places it is difficult to detect the edge that forms the boundary between Fi, the background, and O. On the other hand, there is often a large difference between the signal level of the reverse Km and the signal level of the background, and there is also a portion somewhere where the edge can be easily detected. However, the conventional method
, the X-coordinate of the starting point of the object image must be determined at the Y-coordinate detection position, and since the location was fixed on the 9th direction, the location may be at a location on the edge where it is difficult to detect, as described above. Tatoki F
This has the disadvantage that the i error increases.

In order to improve this drawback, a device for determining a reference position within one AM plane has been proposed in Japanese Patent Application No. 195671 of 1982.

According to this, instead of setting the coordinates X and Y of the point representing the reference position of the object as the X and Y coordinates corresponding to the same point on the image, they are separated and each position is determined with less error depending on the shape. The xis and Y coordinates are determined separately aK, and the determination of the Y coordinate exceeds the required period length in the X direction l scanning period, and the predetermined #! When an image signal level exceeding the threshold level of 1 is detected, 6 Y of the scan is detected.
l! The X coordinate is determined by selecting a location with little error and a location where edges can be easily detected, and determining the X coordinate at a point where the imaging signal during scanning in the X direction exceeds the second threshold level.
This is done using the X coordinate of the position.

The principle of the image reference position determination device proposed above will be explained in more detail.

Figure 4 fm, water whose Y coordinate in the third peng is Yl ″v
This is a waveform diagram of the 2H output of the imaging signal obtained when scanning along the 2-in Ll.
Photograph gI obtained when scanning along the horizontal fin (including 8 points and b5 points) where * is Yl! This is the waveform of the binary output of the signal. In FIG. 3, the starting point 8 for which the Y coordinate is to be determined is determined by scanning along a horizontal scanning line that includes the 8 points. Let us consider that the point where Pfr exceeds the fixed threshold value.

In this way, as can be seen from FIG. 4(a), there is no possibility that the noise 3 or the protruding portions a2-b2 of the image in the first third are mistakenly determined as the starting point. The above is the method for determining the Y coordinate according to the above proposal.

Next, please refer to Figure 7. Now, assuming that the object g17 has a shape as shown, in the light leakage region 81, the Y coordinate can be determined as Y, for example, by using the method described above, but the X coordinate has a large error. What you don't want is
It has already been detailed with reference to No. 5 mA. If we look at the shape of object 7, we can see that in the 'fic&2 area, which is separated by YL from the S1 area in terms of Y coordinate, the X
The coordinates are constant and XI without error! Sometimes it is possible to find the M mark. Therefore, if we calculate the coordinate Xt of the point where the rising edge of the imaging signal obtained by placing $P in the 82 area and performing horizontal scanning) and exceeds a predetermined threshold level, and use it as the X coordinate, we can obtain a value with less error. can be obtained. Moreover, if the threshold level used when determining the Y coordinate and the threshold level used when determining the X coordinate can be determined separately according to the actual signal level of each area. , a more appropriate coordinate position can be determined.The present invention has been made to further increase the 11w degree of the reference position determining device according to the above-mentioned proposal. An object of the present invention is to provide a method for determining a reference position in an X-Y two-dimensional image that is more accurate than the conventional method.

The main point of the configuration of the present invention is that the image 1# that appears on the screen! Select a location suitable for determining the Y coordinate of the reference position and a location suitable for determining the X coordinate based on the shape of 2
@[It is at the point where the coordinates are defined using the conversion signal.

The principle of the present invention will be explained in detail with reference to Figure 8.

The eighth figure is a perspective view similar to FIG. 7 for explaining the operating principle of the present invention. Wind area W2 (a, b r cr rd) at a location suitable for determining the Y coordinate, and wind area W1 at a location suitable for determining the X coordinate.
(., f, g, h) are set as indicated.

In the window area W2, the image 7 is
Segment length of the horizontal scanning output obtained by scanning in the axial direction (for example, N at the Y coordinate position YSY)tut
If the NO value exceeds a predetermined threshold value at any previous Y coordinate position, then the NO value exceeds the predetermined threshold value for the first time, for example, continuously. If the threshold value is exceeded at four coordinate positions, then the fourth coordinate position (if YSY, YS
Y) is assumed to be the YJi image located at 131 (in this case, the Y coordinate position is assumed to advance from bottom to top in the window area W2).

The reason for ``four'' and ``9'' is not limited to this, but it is because it is more resistant to noise if there is only one number. The threshold value may be determined arbitrarily and appropriately depending on the shape of the object.

Assume that the Y coordinate position YSY of the reference position is established in this way. Then, at the Y coordinate position YDX, which is a preset distance @N8 (N8 can be added arbitrarily depending on the shape of the image, etc.) along the Y axis from the Y coordinate position YSY. , examine the output within the window area Wl of the horizontal scanning output obtained by scanning in the X-axis direction,
Find the point at which it rises (Nk of gI), mt. For example, the window area Wl is centered on the Y coordinate position @YDX on eight scanning lines each in the vertical and horizontal directions.
Find the %Xjiilt of the above rising point in
mx oak xsx) is defined as the X coordinate of the reference position. The reason for taking the average is that doing this will improve the degree by 1 cm. Picture 111! If there are no eclipses, the number of eclipses to be averaged will be reduced accordingly. In this way, the reference position
Coordinates and Y coordinates are defined.

Window area W1. Depending on how W2 is taken, it is possible to select a suitable arbitrary location in the image and define the coordinates of the 1st quasi-position at a high n degree.

It is believed that the operating principle of the present invention has been clarified above.

WJ9Y is a block diagram showing one embodiment of the present invention.

In the same figure, 13 is a comparator for Fi2m[, 14 is a 4ImiIL filter, 15 is a counter, and 16
1 is a latch circuit, 17 is a memory, 18 is a change point detector, 1
9 is a latch circuit, 20 is a memory, 21 is a CPU, 22F
i window area generation section, 23ii' pixel division circuit.

Next, the operation will be explained. An analog signal from an iTV camera (not shown) is input to the comparator 13, converted into a digital signal B according to an appropriately set threshold value, and is converted into a digital signal B via the circuit 23. @Input to elementary filter 4, and C is output. 4
A one-field cumulative filter is a digital filter that outputs only when four pixels are continuously at a high level. By providing such a filter, it is possible to remove noise composed of four or fewer pixels.

Using this output C from the 4'm element filter 14, determine the Y coordinate ysy and the x coordinate X of the reference position in the subject (image).
Discover SX. First, the signal C output from the 4-pixel filter 14 is input to the counter 15 and the change point detection section 18, respectively. The counter 15 is supplied with a signal for creating the window W2 shown in FIG.

As mentioned earlier, this window W2 Fi can be set at a position where an image can be taken more accurately. The number of strokes of the image in front of the window 2 is counted by the counter 15 for each scan line, and the fil is latched by the latch 16. This latched data is input to the memory 17 at the end of the scan line. After scanning the 11 images, the CPU 21 uses this data to determine the Y coordinate Y8Y of the reference position according to the W principle described above. In addition, the change point detection unit 18 also has a window Wl shown in Figure @8.
A signal for creating ii! in the window Wl is supplied from the window area generator 22, and ii! Only the image signal ii is input to the change point detection s18, and the image gl within this window Wl
By scanning p for each physical scan line, change point detection unit 1
8, the X coordinate of the point of change between white and black is determined and held by the latch 19, and written to the memory IO for each scanning line. After scanning one screen, the CPU 21 uses this data to determine the reference point according to the principle explained earlier. Position X @ XS
Find X.

As explained above, the BAIC of the present invention has the advantage that the reference position of the image to be inspected can be defined to a high degree by providing a window area.

Note that the present invention is applicable to the case where a product consisting of a moving object is inspected with an iTV camera, since positioning of the reference position of the liquid object in the iTV camera image is necessary for inspecting the subject by pattern recognition etc. .

[Brief explanation of drawings]

Fig. 1 is an explanatory diagram of the captured image screen, Fig. 2 is a waveform diagram showing the waveform of the imaging signal and its binarized result, and Fig. 3 is an explanatory diagram showing -fIlt- of the captured image including noise. Figure 4 is the third
Waveform diagram of the binarized signal of the camera output for the screen shown in Figure 5.
6 and 6 are explanatory diagrams showing the shape of the object image necessary to explain the shortcomings of the prior art, respectively, and FIG. FIG. 8 is an explanatory diagram showing the shape of the object, FIG. 8 is an explanatory diagram similar to FIG. 7 for explaining the operating principle of the present invention, and FIG. 9 is a block diagram showing an embodiment of the present invention. . Code explanation 13... Comparator, 14... 4 pixel filter,
15...Counter, 16.19...Latch circuit, 1
7゜20...Memory, 18...Change point detection unit, 21
. . . CPU, 22 . . . Window area generation unit, 23.
...Pixel division circuit agent Patent attorney Akio Namiki Agent Patent attorney Akira Matsuzaki 1st museum 3! l Figure 2 84 Figure S b3 FA I'l-Otori 62-4 Figure 5 Figure 6 Aχ 11117 Figure JP-A-5858-190708 (5) Figure

Claims (1)

[Claims] 1) An imaging signal obtained by scanning a coordinate signal of a reference position on an imaging screen represented by an X coordinate and a Y coordinate that are perpendicular to the rhinoceros, along said coordinates. 2
A first window that is a method for determining a reference position in an X-72-dimensional image obtained from the value output and output, and that includes a location suitable for determining the Y coordinate of the reference position when viewed from the 1 meter shape. step 1' of establishing a second window area in the image plane, which includes a location suitable for determining the X coordinate as well as the area; means for measuring the segment length for each scanning line and defining the Y coordinate of the reference position based on the measured segment length;
The X coordinate of the reference position M is calculated from the change point of the output of the second winding, which is obtained by scanning the second window area in the t-x axis direction at the Y coordinate position that is a certain distance away from the coordinates. and a means for defining X-Y2.
Reference position determination method in dimensional images.