JPH03123808A - Device for detecting defect of surface - Google Patents

Abstract

PURPOSE:To remove the ring part of a work and the ruggedness of a textured surface as a noise from an image for identification by illuminating the surface of the work from different directions and performing image processing by the logical operation of the change of binary signal to a picture element output before and after changing an irradiating position concerning a non-plane part where there is the difference of height. CONSTITUTION:A video output from a CCD camera 3 is converted to a binary image in a pre-processing part 6. The binary image whose picture element output exceeds a fixed value, for example, is displayed as a white image and the binary image whose picture element output does not satisfy the fixed value is displayed as a black image. At the stage of the conversion of binarization, both of a part which is along the side of the work on an illumination side and a fine projecting part on the surface of the work other than a defective part in right and left illuminated images are displayed as a white pattern in a black texture. Meanwhile, since the part other than the defective part is removed as the noise by the logical operation by a CPU 7, only defective part nearly remains as the white pattern in the image after the conversion. The area of the white pattern is measured by an identification part 8 and compared with a reference value. Then, the propriety is finally decided based on whether the measured value exceeds the reference value.

Description

Detailed Description of the Invention [Industrial Application Field] The present invention is a method for inspecting a defective part that exists in a so-called matte surface that has minute irregularities, such as the surface of a lead frame after metal vapor deposition. This invention relates to a detection device.

[Conventional technology]

For example, as shown in Toshiba Review (1989, Vol. 44, No. 5, P426~), a workpiece surface defect detection device illuminates the upper and lower surfaces of the workpiece at the same time and detects the surface using a CCD camera. There is a method that detects defective parts by measuring the brightness of the light.

There is also a detection device that extracts an image of a defective part by comparing a standard image with no defects and an actual workpiece image.

[Problem to be solved by the invention]

The conventional device as described above cannot be used to detect planar defects different from the satin surface existing in a normal satin surface, such as abrasions, scratches, and adhering foreign matter.

This is because with the former device, which considers sudden changes in brightness as defective, both convex parts and defective parts of the satin surface appear as white spots, and accurate automatic identification of the two cannot be expected.

The same can be said of the latter device that detects defects based on the presence or absence of difference images. In other words, when inspecting a satin surface, even if there are no defective areas on the surface, white spots due to irregularities in minute asperities remain in the image after processing, and therefore white spots in the difference image are caused by surface defects. It is not possible to determine whether

Furthermore, since the boundaries of the lead frame pattern itself need to be automatically identified as defective parts, images of the boundaries must be removed in the same way, leaving an image of only the defective parts.

Due to the second reason, surface inspection of lead frames and the like after metal deposition is done by visual inspection using a microscope, but this method does not improve work efficiency.

An object of the present invention is to provide a surface defect detection device that is an effective solution to the above problems.

[Means to solve the problem]

The surface defect detection device of the present invention is a device for detecting a planar defect existing on the surface of a workpiece having satin-like unevenness. A preprocessing unit that converts the image into an image, a plurality of light sources placed around the workpiece, an arithmetic processing unit that takes in a plurality of binary images and converts it into an image for identification, and a conversion image identification unit. have. In this device, a plurality of light sources are turned on exclusively to illuminate the work surface from diagonally upward and diagonal positions. Further, the arithmetic processing section performs a logical operation on the binary image obtained when each light source is turned on, and deletes pixels whose binary signals have changed due to fluctuations in the irradiation direction from the identification image as noise.

That is, pixels determined to be noise are made the same color as the background to create an image in which only defective parts are extracted. Then, the identification unit measures the area of the residual image after conversion, and if the value exceeds a reference value, it is determined that there is abnormal growth on the workpiece surface.

[Effect]

When the work surface is illuminated from different directions, the pixel output before and after changing the irradiation position will change with a change in the binary signal for non-flat areas with height differences. It is possible to remove the unevenness of the part and the satin surface from the identification image as noise.

Figures 3 to 5 explain the principle in an easy-to-understand manner.

First, regarding the contour of the workpiece, due to the effect of diagonally upward illumination, in the left illumination image of Fig. 3 Ca), the slope part 22 on the left side of the workpiece pattern 20 shines brightly, but the other parts are dark (observable). However, in the right illumination image in the same figure (bl), the high brightness area moves to the sloped area on the right side of the workpiece.In this way, the position of the shadow can be adjusted by changing the illumination direction for areas with steps in the height direction. will change.

The same thing can be said about the illumination image of the convex portion 24 in 21 shown in FIG. 4 as shown in FIG.
, the surface 25 appears bright and unchanged in both left and right illumination.

Therefore, if we consider that the signal after binarization of the bright part is 1 and the dark part is 0, in Figs. 3 and 4, (a) and (b)
), the entire image becomes 0, and the image after processing (C)
The uneven portion is removed from the surface. On the other hand, in Figure 5 (a)
When the AND of (b) is taken, the defective part becomes 1 and the others become 0, allowing the defective part to be extracted. Based on this principle, it is possible to extract only planar defects existing in areas with satin-like unevenness.

Also in the case of logical sum, the relationship between the above signals 1 and O is reversed and Ol
If it is set to l, the logical sum of the defective part becomes I, and all others become O, and the defective part can be extracted.

In this way, the algorithm of the device of this invention extracts only the planar defective part in the pear-shaped surface and compares its area with the reference value to determine whether it is good or bad, making it possible to perform high-precision inspection of the pear-shaped surface. Become.

〔Example〕

FIGS. 1 and 2 show an embodiment of the apparatus of the present invention.

Reference numeral 20 in FIG. 1 is a workpiece to be inspected, for example, a lead frame after aluminum vapor deposition. 2 is a microscope facing the top surface of the workpiece, 3 is a camera connected to 2, and 4.5 is an illumination light placed 180° to the left and right of the center of the camera's field of view. Note that the camera 3 in the figure is a CCD camera using a CCD area sensor 3a as a solid-state imaging device.

FIG. 2 is a block diagram showing the entire detection device 1 of the present invention. As can be seen from the figure, the video output from the CCD camera 3 is converted into a binary image by the preprocessing section 6. In the binary image, those whose pixel output exceeds a certain value (pixels that capture high-brightness areas) are displayed as white pixels, and those whose pixel output is less than a certain value (pixels that capture low-brightness areas) are displayed as black pixels. be done. In addition, in both the left and right illumination images, in addition to the defective parts, the parts along the side of the workpiece on the illumination side and minute convexities on the workpiece surface appear as self-patterns in the black background. However, for parts where there is a change in the binary signal in the left and right illumination images, that is, parts other than defective parts, we will use C after this.
PU? is removed as noise by a logical operation, so only the 1' defective part remains as a self-pattern in the converted image. The area of the remaining self-pattern is measured by the identification section 8, and compared with a reference value, and a final judgment is made as to whether it is good or bad based on whether it exceeds the reference value.

Note that it is desirable that the lights be placed at angles that divide a circle centered on the work into N equal parts. In addition, if the lights are fixed, the number of lights to be installed is small (two lights are installed in each case, but if the structure is such that the workpiece is illuminated at a fixed point while moving around the workpiece with a rotating mechanism, only one light is required). Good. Although the latter method is mechanically more complex, the irradiation direction can be changed and the irradiation position can be increased without limit.

In addition, the microscope was installed because it is difficult to treat as noise when there are extremely fine unevenness on the lead frame surface, ranging from several microns to tens of microns, unless it is magnified at a considerable magnification. can be omitted.

[Effects] As described above, in the detection device of the present invention, images caused by steps on the outline of the workpiece or unevenness on the surface are removed as noise from the final identification image. It is possible to automatically detect flat defects in the satin surface with high accuracy, which could not be detected with IC.
It can contribute to improving the efficiency of surface inspection of IJ-do frames, etc., and improving reliability.

[Brief explanation of drawings]

FIG. 1 is a perspective view showing an outline of an example of the device of the present invention;
Figure 2 is the same block diagram as above, Figure 3 (a), 0)), (
C), Figure 4 (a), et al.), (C), Figure 5 (a), 0
)), (C) are explanatory diagrams of the detection principle, and (a) (
1) shows the irradiation direction, (n) shows the image, and (III) shows the state of light reflection. Also, (c) (1) is the irradiation direction,
(II) shows the image, and (C) shows the image after processing. 1...Detection device, 2...Microscope, 3...CCD camera, 4.5...
・Light, 6...Pre-processing section, 7...
...CPU, 8...Identification unit. Same Agent Aya Kamata Figure 3 22 ■ ■ ■ (C) -Yu Illumination Light 11111 Light ■ ■ ■ (C) Figure 5 (b) (C)

Claims (3)

[Claims] (1) A device for detecting a planar appearance defect existing on the normal surface of a workpiece with satin-like unevenness, which includes an image input device that captures the surface of the workpiece, and converts the input image into a binary image. a plurality of light sources disposed around the workpiece, an arithmetic processing section that captures a plurality of binary images and converts them into images for identification, and a converted image identification section, The above-mentioned light sources are turned on exclusively to illuminate the work surface from diagonally upward anisotropic positions, and the above-mentioned processing section performs logical operations on the binary images obtained when each light source is turned on, and generates a binary signal by changing the irradiation direction. The changed pixels are deleted from the identification image as noise, and the identification section measures the area of the residual image after conversion, and if the value exceeds a reference value, it is determined that there is an abnormality on the workpiece surface. Surface defect detection device. (2) The surface defect according to claim (1), wherein a single light source orbiting around the workpiece is used instead of the plurality of light sources, and the position of the light source is changed to illuminate the workpiece surface from anisotropic angles. Detection device. (3) Claim (1) in which the image input section is equipped with a microscope.
Or the surface defect detection device described in (2).