JPH0115909B2 - - Google Patents

Description

[Detailed description of the invention]

The present invention relates to an inspection device for inspecting the quality of an object to be inspected, such as a bottle or a container, whose cross section has an axially symmetrical shape such as a circle or an ellipse. In the past, such inspections were performed exclusively by visual inspection, but this had problems such as securing inspection personnel, instability of inspection standards, and cost.Therefore, recently, various automatic inspections have been used to replace visual inspections. Proposed. Examples include a "method for determining shape coefficients" and a "contour coordinate matching method." The former involves capturing an image of the object to be inspected using, for example, an industrial television camera (hereinafter also simply referred to as a TV camera), converting the video signal into a binary value at a predetermined threshold level and storing it in a memory. , based on the stored data, calculate the area S, perimeter L, and shape coefficient k (=S/L ² ) of the object to be inspected. For example, in the case of a "circle", the shape coefficient k will be a predetermined value. This is to determine the roundness of the inspection object. However, although the calculation speed is relatively high according to this method, it has a serious drawback that the coefficient k does not change much even if the object is slightly deformed, that is, the detection sensitivity or accuracy is poor. On the other hand, in the latter method, the shape of the object is inspected by determining the coordinate points forming the outline of the object image from the information stored in the same manner as described above, and comparing these points with standard data. Therefore, the detection accuracy for deformation is considerably higher than the former, but if the dimensions of the object differ even slightly, it will no longer match the standard pattern, so matching must be performed many times, resulting in a significant increase in calculation time. It has the disadvantage of being long. Furthermore, since a large number of standard pattern tables are required, the memory capacity becomes large, resulting in an increase in cost. The present invention has been made under these circumstances, and an object of the present invention is to provide an inspection device that can inspect the quality of an object having an axially symmetrical shape such as a circle at high speed, with high precision, and at low cost. With the goal. Its feature is that it binarizes the imaging signal obtained by imaging the axially symmetrical inspection pattern by aligning the symmetry axis with the vertical scanning direction of the horizontal (X) and vertical (Y) scanning imaging device. Detect the minimum X coordinate and maximum X coordinate for each horizontal scanning line from the rising and falling points of the binarized signal, store these, and from the stored data, detect the center point of the inspection pattern and pass through the center point. Find a center line parallel to the Y axis, calculate each deviation of the minimum and maximum X coordinates from the X coordinate of the center line, determine the symmetry of the inspection pattern from the deviation, and calculate the X, Y By calculating each maximum outer length (maximum diameter) in the axial direction and its deviation, and determining the outer shape of the test pattern from the deviation, the acceptability of the test pattern can be detected quickly and with high precision. be. Furthermore,
The inspection pattern is divided into a plurality of regions along its central axis, and the deviation and polarity between the minimum and maximum X coordinates on each adjacent horizontal scanning line within each region are detected, and the polarity is calculated based on the detection results. By adding a function to determine the outline of the inspection pattern, comprehensive inspection is performed and detection accuracy is further improved. Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing an embodiment of this invention.
FIG. 2 is a block diagram showing another embodiment of the comprehensive judgment circuit shown in FIG. 1, and FIGS. 3 to 7 are explanatory diagrams illustrating a binarized image of a pattern to be inspected. In FIG. 1, 1 is a TV camera, 2 is a binarization circuit, 3 is a screen division circuit, 4 is a mask generation circuit,
5 is an AND gate, 6 is a minimum X coordinate detection circuit, 7
8 is a maximum X coordinate detection circuit, 8 is a memory, and 9 is a comprehensive judgment circuit consisting of an outermost point detection circuit 10, a center point detection circuit 11, a symmetry judgment circuit 12, an outer dimension judgment circuit 13, and an OR gate 14. Here, the operation will be explained with reference to a pattern in which the inspection target pattern is represented by a "circle" as shown in FIG. 3A. A video (imaging) signal obtained by capturing an image of the inspection object with the TV camera 1 is binarized in the binarization circuit 2, and is converted into horizontal (X),
It is discretized (pixelized, picture elementized) into a large number of vertical (Y) two-dimensional grid points. Note that this discretization uses a high frequency clock (6MHz) for the binary signal in the horizontal (X) direction, and a high frequency clock (6MHz) in the vertical (Y) direction.
This is easily done by using each horizontal scanning line of the TV camera 1. The video signal discretized or pixelized in this way is output by the mask area generation circuit 4 and the AND gate 5 to the minimum X coordinate detection circuit 6 and maximum X coordinate detection circuit 7 in which only information regarding the desired area of the imaging screen is output be done. Note that the mask area is designated as a horizontal and vertical two-dimensional area. The minimum X coordinate detection circuit 6 detects the coordinates of the first rising point of the binarized signal for each horizontal scan, that is, for each arbitrary Y coordinate yi as shown in FIG. 3A.
x _pi and the maximum X coordinate detection circuit 7 detects the last falling point coordinate x _ni , respectively, and stores them in the memory 8. This operation is performed for all horizontal scanning lines forming one screen or for all horizontal scanning lines within the area specified by the mask generation circuit 4,
As a result, the x _pi area of memory 8 stores the X coordinate of the point indicated by the white circle mark (〇) in Figure 3A, and the x _ni area stores the X coordinate of the point indicated by the black circle mark (●). Ru. Here, if the memory address is made to correspond to the Y coordinate and essentially only the X coordinate is stored, the memory capacity can be reduced, and access to it becomes easier, resulting in faster calculation speed. It is something. The information stored in the memory 8 in this manner is read out by an arithmetic and control unit (not shown) and provided to the comprehensive determination circuit 9. In the outermost point detection circuit 10 of the comprehensive judgment circuit 9, the minimum value x _pp and maximum value x _no of the X coordinate within the effective screen, and the minimum value of the Y coordinate
y _pp and the maximum value y _no are detected, and the maximum outline length in the X and Y directions of the target pattern figure D _x (x _no −x _pp ), D _y
(y _no −y _pp ) is obtained. In addition, in Fig. 3A, P _p (o, o) is the origin, P ₁ and P ₂ are the minimum and maximum X
The coordinate generation points L _p and L _n indicate the minimum and maximum Y coordinate generation lines, respectively. The output from this outermost point detection circuit 10 is given to a center point detection circuit 11, and the center point detection circuit 11 determines its center coordinates P _c (x, y). This center coordinate P _{c
One way to find the}
The average value of x _pi and x _ni for the region (α: constant) can be found and this can be set as the center of the target pattern figure in the X-axis direction. Expressing this in the formula, becomes. Further, the center P _c (y) in the Y direction can also be found in the same manner as above, but in this case, it is also possible to simply replace y' _c with P _c (y). The symmetry determination circuit 12 uses a left-hand line whose central axis is an imaginary line A-A' (the _center line of the , the length of each X axis to both right ends is determined, and its symmetry is determined by checking whether these are within a predetermined threshold. In other words, the following calculation is performed for each horizontal scanning line _yi : x _ni +x _pi -2P _c (x) (1). At this time, if the object to be inspected is a good product (a perfect circle), the value of equation (1) above will be "0" or approximately "0", and if it is a defective product, it will be greater than a predetermined value (>β), so it is good. , it is possible to determine whether the product is defective or not. Note that it may be possible to determine a defect when the value of equation (1) exceeds a predetermined value (β) for any one of the horizontal scanning lines, but since there is also the influence of noise etc. It is more appropriate to determine that the product is defective when the condition is satisfied in the horizontal scanning lines (n rasters) (selected as =2 to 3). Since the symmetry determination circuit 12 outputs, for example, a logic "0" signal when the product is good, and a logic "1" signal when the product is defective, at a predetermined timing, the output is further given to a required device via an OR gate 14. By the way, as shown in Fig. 4, there is a defective part due to a bulge in a part approximately on the center line of the object to be inspected.
F ₁ may exist. In this case, since the above-mentioned symmetry determination alone is insufficient, as a countermeasure, the following calculation is performed by the outer dimension determination circuit 13. That is, using the data D _x and D _y obtained by the outermost point detection circuit 10, the following calculation is performed: |D _x −D _y | (2). If the product is good, the value of equation (2) is within the predetermined value (γ), and if it is defective, it is greater than the predetermined value (γ), so it is possible to determine whether the shape of the inspection pattern is good or not. . Note that if you normalize by performing the operation |(D _x −D _y )/D _x | ...(2)′ instead of formula (2), it will be applied not only to circles but also to shapes such as ellipses. Not only can this be done, but the algorithm can also be made so that the influence of size fluctuations can be ignored. This is because the value of the above equation (2)' is a constant value unique to a pattern such as a circle or an ellipse. In addition, the above
(2)′ Maximum diameter in the X-axis direction (maximum outer length) D _x
Instead of standardizing by
Of course, it may be normalized by D _y .
In this way, from the OR gate 14, the symmetry judgment circuit 1
2. Since the OR output from the outer dimension determination circuit 13 is obtained, even if at least one of the determination results is defective, the pattern to be inspected is determined to be defective. For example, as shown in FIG. 3B, the "beard" part F ₂ ,
It can be seen that the target pattern having a defective portion such as the "dent" portion _F3 is determined to be defective by both the symmetry determining circuit 12 and the outer dimension determining circuit 13. Therefore, there is a bulge _F1 as shown in Figure 5A, and a "beard" as shown in Figure 5B.
F ₂ , a dent F ₃ , or a "burr" F ₄ as shown in Figure C, the quality can be determined by determining symmetry or outer dimensions. However, it can be seen that the above determination method cannot distinguish, for example, between a circle and a square, or between a circle and a regular polygon, due to its characteristics. Therefore, in the present invention, an increase/decrease determination circuit 15 is further added as shown in FIG. The increase/decrease determination circuit 15 performs the following calculation on the valid Y-axis area where the test pattern exists from the data in the memory 8. That is, focusing on arbitrary Y coordinates y _i and y _i+1 , the minimum X coordinates at y _i and y _i+1 are x _pi , x _p(i+1) , and the maximum X coordinates are x _ni , x When _n(i+1), calculate x _pi −x _p(i+1) = DF _pi x _ni −x _n(i+1) = DF _ni ...(3), and calculate the calculated value DF _pi , Examine the absolute value and polarity of DF _ni . In other words, for example, if the object to be inspected is a square, DF _pi obtained by the above equation (3),
DF _ni is a constant value regardless of the yi coordinate, but if the object is circular or elliptical, it changes depending on the yi coordinate.
D _y obtained in , for example _, is _divided into eight equal parts as shown in FIG. ~
Split into A8. The absolute values n ₁ to n ₈ and n' ₁ to n' ₈ of the above DF _pi and DF _ni in these regions A1 to A and their polarities are examined as shown in the following table. Therefore,

[Table] By preparing a judgment table as shown in the above table in advance in the increase/decrease judgment circuit 15, the absolute values of DF _pi and DF _ni (n ₁ to n ₈ , n′ ₁
~n' ₈ ) and its polarity (+, -) can be compared using a table. Furthermore, when the object is a circle, the above n ₁ = n ₈ = n' ₁ = n' ₈ , n ₂ = n ₇ = n' ₂ =
It can be seen that n' ₇ , n ₃ = n ₆ = n' ₃ = n' ₆ , n ₄ = n ₅ = n' ₄ = n' ₅ . In this way, the increase/decrease judgment circuit 15 outputs a logic "0" indicating that the object to be inspected is a good product when both the absolute value and polarity pass, and outputs a logic "1" indicating that the object is defective when at least one of them fails. do. Therefore, if the overall judgment circuit is configured as shown in FIG. 2, even more accurate judgment can be made. In FIG. 2, if a defective signal is output from at least one of the symmetry determining circuit 12, the outer dimension determining circuit 13, and the increase/decrease determining circuit 15 via the OR gate 14, it is determined that the object to be inspected is defective. Needless to say, it will happen. Furthermore, for a pattern consisting of double closed loops C ₁ and C ₂ as shown in FIG. 7A, the mask generation circuit 4 in FIG. , it is possible to determine whether the pattern is good or bad by generating a mask for distributing it to the outside, separating it into separate patterns C ₁ and C ₂ as shown in B and C in the same figure, and making the same judgment as above for each pattern. It is something. Above, we have mainly explained the case where the object pattern is circular, but if it is axially symmetrical, appropriate setting values may be selected for ellipses, rectangles (including squares), quadrilaterals, or other shapes.
Alternatively, the acceptability of the shape can be determined by applying a standardized determination algorithm. As described above, according to the present invention, a video signal obtained by imaging an object to be inspected with a two-dimensional scanning imaging device is binarized and stored in a memory, and the center of the object is determined from the stored image data. Since the symmetry is determined using the center line as the basic axis, abnormalities in the outer shape of the object (concavity, convexity) can be detected with high precision and stably regardless of the size of the object. Can be done. In other words, the symmetry of a figure is unrelated to its "size." Furthermore, since the reference axis is aligned with the Y axis, that is, the vertical scanning direction of the imaging device, and inspection is performed focusing only on the X coordinate, the hardware can be simplified and calculations can be performed at high speed.
By using the minimum and maximum values of the X coordinate within the effective area as target data for the determination process, the pattern is enlarged, especially for whisker-like external protrusions, that is, the external protrusions are exaggerated. Detection sensitivity increases. In addition, by performing outer dimension judgment in addition to symmetry judgment, it is possible to easily and quickly detect patterns such as the one shown in Figure 4, which cannot be detected by judging symmetry alone, and with high precision regardless of size. can be determined. Note that these determinations only need to be made once regardless of the size of the pattern, so high-speed processing is possible. Furthermore, symmetry,
By using outer dimension judgment together, it is possible to detect not only biaxially symmetrical figures such as circles, but also uniaxially symmetrical figures with exactly the same accuracy and at high speed. It has the great advantage of requiring only changes. In addition, by using an increase/decrease judgment circuit to detect local changes in shapes such as unevenness, shapes that cannot be detected by "symmetry + outer dimension" judgment (for example, the difference between a circle and a square) can be detected even if the size is small. Even if there is a change, it can be detected with high accuracy with a single calculation. Also,
Divide the Y-axis contour data of the figure obtained by the increase/decrease judgment circuit into n regions and detect changes in polarity and absolute value for each region, thereby distinguishing and detecting circles and polygons, for example. Can be done.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an embodiment of this invention.
FIG. 2 is a block diagram showing another embodiment of the comprehensive judgment circuit, and FIGS. 3 to 7 are explanatory diagrams illustrating a binarized image of a pattern to be inspected. Description of symbols, 1...TV camera, 2...Binarization circuit, 3...Screen division circuit, 4...Mask generation circuit, 5...And gate, 6...Minimum X coordinate detection circuit, 7...Maximum X coordinate detection circuit, 8...Memory, 9...Comprehensive judgment circuit, 10...Outermost point detection circuit, 11...Center point detection circuit, 12...Symmetry judgment circuit, 13...Outline dimension judgment circuit, 14...
OR gate, 15...Increase/decrease judgment circuit.

Claims (1)

[Claims] 1. Horizontal (X) and vertical (Y) scanning type imaging means for capturing an image of an axially symmetrical inspection object by aligning the axis of symmetry with its vertical scanning direction; binarization means for binarizing the captured image signal at a predetermined threshold level, and a minimum X coordinate and a maximum X coordinate, respectively, from the rising edge and falling edge of each horizontal scanning line of the binary image capturing signal. Minimum and maximum X coordinates detection means for detecting, storage means for storing the minimum and maximum X coordinates for each detected horizontal scanning line, and a Y axis passing through the center of the inspection target pattern from the stored data. symmetry determining means for determining the symmetry of the inspection pattern based on the deviations from the center line X coordinate and the minimum and maximum X coordinates; , an outer dimension determining means for calculating each maximum outer length (maximum diameter) in the Y-axis direction and a deviation thereof and determining the shape of the inspection pattern from the deviation, at least the symmetry determining means and the outer dimension determining means. A pattern inspection device characterized in that the quality of the shape of an object to be inspected is determined from one output. 2. The pattern inspection apparatus according to claim 1, wherein the symmetry determining means determines that the inspection pattern is defective when the symmetry of the inspection pattern does not hold over at least two horizontal scanning lines. Device. 3. In the pattern inspection device according to claim 1, the outer dimension determining means normalizes the deviation of each maximum outer length in the X and Y axis directions using the maximum outer length in either axis. A pattern inspection device featuring: 4. In the pattern inspection device according to claim 1, the inspection pattern is divided into a predetermined number of regions based on the stored data, and each minimum X coordinate on each horizontal scanning line adjacent to each other within each region is divided into a predetermined number of regions. 1. A pattern inspection device comprising a contour determining means for determining the contour of an inspection pattern by detecting deviations between the test patterns and the maximum X coordinates and their polarities, and referring to the determination results. 5. In the pattern inspection apparatus according to claims 1 to 4, the inspection patterns consisting of a plurality of closed loops are separated by generating a mask of a predetermined shape, and each separated pattern is inspected for symmetry and outline. A pattern inspection device characterized by determining dimensions or contours.