JPH03252544A - Wiring pattern inspection device - Google Patents

Abstract

PURPOSE:To perform inspection without any erroneous information even if wiring patterns differing in line width are present thereon by converting binary image data on a printed board into a distance image and making a selection with plural width threshold values. CONSTITUTION:A wiring pattern formed on the printed board 101 is converted photoelectrically by an image input means 102 and the obtained a gradation image is subjected to binarization and converted 105 into a binary image. This image is converted by a distance image converting means 106 into a distance image indicating the distance of the wiring pattern from the background. Then the distance value of a ridge point in the distance image is compared with an optional set threshold value to detect a flaw in wiring pattern width. Thus, the pattern is converted into the distance image to recognize the distance value, i.e. wiring pattern width directly, thereby easily performing the inspection.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of the Invention The present invention relates to a wiring pattern inspection device for inspecting wiring patterns on printed circuit boards, photomasks, etc. for defects.

BACKGROUND OF THE INVENTION As the density of mounting electronic components on printed circuit boards increases, wiring patterns are becoming increasingly finer. Conventionally, defects in printed circuit boards, etc. have been visually inspected by humans, but due to the miniaturization of wiring patterns, it has become difficult to continue inspection work for long periods of time while maintaining inspection accuracy. Automation is desired.

As defect inspection methods for wiring patterns, JL, C, Santsyany, K, and J.

L, C, 5anz and A, ni, Jain, ”M
achine vision techniques f
or 1n-spection of printed
wiring boards and thick-f
ilm circuits.

“J, Opt, Soc, Amer, , vol.
3. noog, pp4465-1482.5ept
, 1986) LaK! D: Many methods have been introduced, and in particular, many methods classified as design rule methods or comparison methods have been proposed. These methods have their advantages and disadvantages, but one of the most promising and interesting methods is Zhu. R, Mandevi
Lie, “Novel Method forana
lysis of printed circuit
images, ” IBMJ, Res, Develo
p,,vol,2g,no,1. pp, 34g-376
There is a method (Jan, 1979) in which binary image data is contracted or expanded and then thinned to detect defects in wiring patterns, which will be explained below as a conventional example. FIG. 8 shows the procedure of defect detection processing.

Figures (a) to (d) show the missing defect detection procedure;
1 to (h) show the protrusion defect detection procedure.

First, a method for detecting missing defects will be explained with reference to the drawings. (a) shows a defect image, where point b and point C are detected as fatal defects due to insufficient line width and disconnection, and point aA is not detected as a defect. As the first step (
In b), the connection at point b is cut off by shrinking (eroding) the image by a predetermined size. In the second step (C), the pattern is thinned to a width of one pixel. In the third step (d), the connectivity of the thinned image is determined in a 3×3 local area (the position indicated by the opening in the figure), and points b and C are detected as disconnections. Note that the connection determination of the 3×3 local area shows that the junction between the terminal portion and the wiring pattern (the position indicated by a circle in the figure) can also be detected as a feature point.

Next, a method for detecting a protruding defect will be explained with reference to the drawings. (e) shows a defect image, where points b and C are detected as fatal defects due to line width abnormalities and shorts, and a
Points shall not be detected as defects. As the first step (f), by expanding the image by a predetermined size, b
Generates a new connection between points. As a second step (g)
The pattern is thinned to Fi1 pixel width. As a third step, in (h), the connectivity of the thinned image is determined in the /ri3X3 local area (the position indicated by the opening in the figure), and points b and C are detected as branch points, that is, shorts. By the above procedure, thick lines, breaks, and shorts can be detected.

In view of the above-mentioned conventional problems, the present invention enables inspection without false alarms even when wiring patterns of multiple line widths exist on the same printed circuit board, and recognizes defects such as disconnections and shorts in addition to line width inspection. The present invention provides a wiring pattern inspection device that enables various defect inspections.

Problems to be Solved by the Invention A method has been described in which a binary image is contracted and expanded to enhance the characteristics of the defect and then thinned, and the defect is detected by connection determination in a 3×3 local area. This method cleverly utilizes wiring pattern design rules and can reliably detect defects, so it can be said to be a promising method.

However, because the characteristics of defects are amplified by shrinkage and expansion processes, it is not possible to classify into insufficient line width and disconnection, or between thick line width and short. Furthermore, if there are multiple settings for the violation line width, multiple image contraction, expansion, and line thinning processing processes are required, which poses problems such as an increased burden on hardware when implementing the system.

In view of the above problems, the present invention enables inspection without false alarms even when wiring patterns of multiple types of line widths exist on the same printed circuit board, and is capable of not only inspecting line widths but also recognizing defects such as disconnections and short circuits. The present invention provides a wiring pattern inspection device that can perform various defect inspections.

Means for Solving the Problems To solve the above problems, the technical solutions of the present invention firstly include an image input means for optically detecting and photoelectrically converting a wiring pattern formed on a printed circuit board; binarization means for converting the gray scale image signal from the input means into a binary image signal; distance both-edge conversion means for converting each pixel of the wiring pattern into a distance image indicating a distance value from the background; The defect detecting means detects a ridge point and compares the distance value of the ridge point with an arbitrary set threshold value to detect a defect in the wiring pattern width.

Second, in addition to the above-mentioned means, the defect detection means compares the distance value of the ridge point with a plurality of arbitrary threshold values to detect defects in a plurality of wiring pattern widths. be.

Thirdly, the defect detection means extracts features such as branches and terminations of the ridge point image, and detects defects such as short circuits and disconnections in the wiring pattern.

Function The present invention first converts a grayscale image obtained by photoelectrically converting a wiring pattern formed on a printed circuit board into a binary image, thereby creating a binary image. Using this binary image, convert it into a distance image that shows the distance of the placement pattern from the background, and compare the distance value of the ridge point in the distance image with an arbitrary set threshold to detect defects in the wiring pattern width. By converting it into a distance image, the distance value, ie, the width of the wiring pattern, can be directly recognized, making it easy to inspect.

Second, by comparing the distance value of the ridge point with a plurality of arbitrarily set values, it is possible to inspect wiring patterns with a plurality of line widths.

Thirdly, various inspections can be performed by extracting features such as branches and terminations from the ridge point image and identifying line width abnormalities, breaks, and sheets in the wiring pattern.

Embodiments Hereinafter, embodiments of the present invention will be described with reference to FIG.

FIG. 1 is a block diagram of a wiring pattern inspection apparatus according to an embodiment of the present invention. In Figure 1, 101
102 is a printed circuit board, 102 is an image input means including a diffused illumination device 104 such as a ring-shaped light guide, and an imaging device 103 such as a CCD camera, 105 is a binarization means for converting a grayscale image into a binary image, 106 107 is a distance image conversion means that converts each pixel of the wiring pattern on the binary image into the shortest distance value from the background; 107 is a distance image converter that uses the distance image to convert the distance value at the ridge point and a certain arbitrary set threshold value; A defect detection means for comparing and detecting defects in wiring pattern width is shown.

The operation of the above configuration will be explained below. FIG. 2 shows a specific example of the configuration of the image input means 102 and the binarization means 105. In FIG. 2, 101 is a printed circuit board, 121 is an xy table, 122 is a white light source such as a halogen lamp, 123 is a ring-shaped light guide for diffused illumination, 124 is an optical lens for imaging, and 125 is a printed circuit board 101. A CCD line sensor that photoelectrically converts reflected light, 126 is a grayscale image signal, 127 is an analog comparator, 1
28 is a terminal for supplying a comparison reference potential to the analog comparator 127, and 129 is an output terminal for a binary signal. The xy table 121 on which the printed circuit board 101 is mounted is
From the imaging start position of 01, it moves in the X direction (sub-scanning direction) at a predetermined pitch, and when scanning is completed, it moves L in the X direction (main-scanning direction). After moving in the X direction, it scans again in the X direction in the opposite direction to the previous scan to scan the entire surface of the printed circuit board 101. The above value is determined by the following formula.

L, , = k・δ・(m-mo) (here, the imaging magnification of the kFi lens, δ is the element size of the CCD line sensor, m is the number of elements of the CCD line sensor,
fpQ indicates the number of pixels of overlap scanning. ) The white light from the halogen lamp 122 passes through the ring-shaped light guide 12.
The reflected light is diffusely irradiated onto the printed circuit board through the lens 124 and sent to the p CCD line sensor 125. The obtained grayscale image signal 126 is sent to an analog comparator 127.
The signal is input to the terminal 128 and is binarized at a predetermined threshold level.

Next, image signal processing in the distance image converting means 106 will be explained with reference to FIG. FIG. 3(a) shows the scanning procedure of a logical mask for conversion into a distance image. For the binary image @131 from the binarization means 105, an M×N scanning window 132 centered on the amateur of interest is formed, and the pixels within the window are scanned pixel by pixel in the main scanning direction while performing logical operations on the pixels within the window.

This procedure is repeated until the end of sub-scanning to obtain a distance image of the same size as the input binary image. The specific configuration of image signal processing using a scanning window is a common technique and its explanation will be omitted. The procedure for converting to a distance image will be explained below. In the same figure (b), M×
FIG. 7 is a configuration diagram of a logical mask when M and N=15 in an N scanning window 132. Pixel numbers 1 to 7 are assigned in order in a circular manner, centering on the amateur on the attention side in the center. The pixel number indicates the distance from the amateur of interest. The distance to the pixel of interest A is determined by the flow shown in FIG. First, in 133, it is determined whether the pixel of interest is 1 or O, and if it is 0, 0 is output as a distance value in 138. If the pixel of interest is 1, the distance value is determined by the steps 134 to 137. In 134, pixel number n is set to 1, and 13
5, the logical AND of the pixels with pixel numbers n is performed. If the logical product is 1, the pixel number n is incremented by 1 at 136, and the determination at 135 is performed again. If the logical product is 0 in 135, n is output as a distance value in 137.

The shortest distance from the pixel position of interest to the boundary of the wiring pattern can be obtained by the above calculation.

Next, a first embodiment of the defect detection means 107 will be described with reference to FIG. FIG. 3A shows a distance image converted by the distance surface edge conversion means 106. '00''
The pixel with the value Q is the base material part of the printed circuit board, and the pixel has the value Q, “0
1"02"03" indicates the distance value of each pixel from the pattern boundary. FIG.
) is scanned using the 5×5 logical mask shown in (b), and when the pixel of interest is not "OO", it is determined whether the position of interest is the ridge position of the distance image using the following equation.

(Di-A, Ds-A) = (negative value, negative value) or (D2-A, D6-A) = (negative value, negative value)
or (Da-A, D7-A) = (negative value, negative value) or (D4-A, Ds-A) = (negative value, negative value) If the above conditions are satisfied, the pixel of interest is determined to be a ridge position, and the line width is determined as follows. When detecting a defect when the margin width is less than W pixels, and assuming that the distance value of the ridge point is A, the defect is determined when the following formula is satisfied.

A≦W/2 For example, when a line width W of less than 4 pixels is determined as a defect, the position of the O mark in (a) is detected as a defect.

As described above, according to this embodiment, by directly comparing the prescribed line width and the distance value at the ridge point of the distance image, defects are continuously detected at the position where the defect exists and at its neighboring positions. line width defects can be detected with simple calculations. In this embodiment, an example of insufficient line width has been described, but a thin line defect can also be easily detected by a similar comparison calculation. When detecting a line width exceeding the specified width W pixels as a thick line defect, it is determined to be a thin line defect if the following conditional expression is satisfied.

A>W/2 Next, a second embodiment of the defect detection means 107 will be described with reference to FIGS. 5 and 6. FIG. 5 shows the result of performing the ridge detection shown in FIG. 4 on the distance image. In the figure, 153 (○ mark in the figure) is the ridge point of the thin wiring pattern 151, 154 (○ mark in the figure) is the ridge point of the thick wiring pattern 152, and! 7.155 and 156
The rectangular area represents a 5×5 logical mask explained in FIG. 6(a). The difference between this embodiment and the first embodiment is as follows:
The point is that, before comparing the distance value of the ridge point and the prescribed line width, a means is provided for selecting and determining a reference value to be compared with the distance value of the pixel of interest from among a plurality of prescribed line widths. The selection and determination of comparison reference values will be explained below with reference to FIG. The same figure (a
) The configuration of a logical mask for selecting and determining H comparison reference values is shown. FIG. 4B is a flowchart showing a procedure for selecting and determining a comparison reference value. First, in step 161, it is determined whether or not the target position A is a ridge point, and if it is not a ridge point, step 162
Then, the process moves on to the determination at the next scanning position. If the target position A is a ridge point, in step 163, the maximum distance value dmax is determined by referring to d1 to d16 of the logical mask shown in (a). Next, in step 164, the threshold value Wth for identifying the line width of 2\iL class given to the device from the outside in advance and the dm
ax are compared and classified into 165 when the comparison reference value is thick and 166 when the comparison reference value is thin. The procedure for line width violation determination in 165 and 166 was shown in the embodiment of 'f41, so the explanation thereof will be omitted. For example, in FIG. 5, the threshold value Wth for identifying the two types of line widths is set to 6 pixels,
When the line width of a thick distributed pattern is less than 6 pixels, it is determined that the line width is insufficient. Even if the target position has the same distance value, the scan position 155 will not be detected as a defect, and the scan position 156 will be detected as a defect. Can be detected as a defect. Note that in this embodiment, line width determination in the case where there are two types of line widths has been described, but if multiple threshold values wth for identifying the line widths are set, inspection of two or more types of line widths is also possible. It is. As described above, according to this embodiment, even when wiring patterns with a plurality of line widths exist on the same printed circuit board, by providing a procedure for selecting and determining comparison reference values, it is possible to conduct an inspection without false alarms.

Next, a third embodiment of the defect detection means 107 will be described with reference to FIG. The difference from the first and second embodiments is that the 5×5 logical mask shown in FIG. 7(a) is scanned against the ridge point image detected by the logical mask shown in FIG. 4(a), and the ridge line The point is that a means for detecting the termination and branching of the is provided. An annular pixel region 172 centered on the pixel of interest 171 is arranged within a 5×5 scanning window, and the end/branch of the ridge line is determined based on the ao-a15 bit pattern. Figures (b) and (c) are examples of terminal and branch detection, respectively. In FIG. 4B, when the pixel of interest 171 is on the ridge line 174, if there is one or less ridge point area in the annular area of the 5×5 logical mask 173, it is determined that the pixel is the end.

In addition, in the same figure (C), the pixel of interest 171 is the ridge line 175.
If there are three or more ridge point regions in the annular region of the 5×5 logical mask 173, it is determined that there is a branch.

The determination of bit patterns aO to a15 in (a) is aO to
This can be easily realized by referring to a lookup table using a15 as an address. In other words, a determination code indicating the termination or branch is stored in advance in a memory cell whose address is the bit pattern of the termination or branch.
Terminations and branches can be detected by referencing the memory using the bit pattern. As described above, according to this embodiment,
By providing means for scanning the ridge point image with an annular mask and determining the bit pattern in the annular region, it is possible to detect terminations and branches, which are graphical characteristics of wire breaks and shorts.

Effects of the Invention As an effect of the present invention, in order to convert the binary data of the printed circuit board into a distance image, the defect detection means directly compares the distance value of the ridge point of the distance image with an arbitrary single-width threshold value. This allows for easy line width inspection. In addition, in the defect detection means, the line width threshold value to be compared with the distance value of the ridge point can be set in multiple stages by selecting the threshold value depending on the scanning position.
Even if wiring patterns with multiple types of line widths exist on the same printed circuit board, inspection can be performed without false alarms. Furthermore, since the defect detection means determines the end/branch of the ridge line, it is possible to recognize not only line width inspection but also defects such as wire breaks and short circuits, making it possible to perform a variety of defect inspections.

[Brief explanation of drawings]

FIG. 1 is a block wiring diagram of a wiring pattern inspection device according to an embodiment of the present invention, FIG. 2 is a detailed block wiring diagram of the image input means and binarization means in the same device, and FIG. 3 is a distance diagram in the same device. FIG. 4 is a diagram showing the signal processing procedure of the one-image conversion means, FIG. 4 is a diagram showing the defect detection procedure of the first embodiment in the same defect detection means, and FIG. 5 is a diagram showing the second embodiment in the same defect detection means. A conceptual diagram of a distance image, FIG. 6 is a diagram showing the defect detection procedure of the second embodiment in the defect detection means,
FIG. 7 is a diagram showing the defect detection procedure of the third embodiment in the defect detection means, and FIG. 8 is a diagram showing the defect detection procedure in the conventional wiring pattern inspection apparatus. 103... Imaging device, 104... Diffusion illumination device, 1
21... xy table, 122... light source, 123...
...Ring-shaped light guide, 124...Optical lens,
126... Grayscale image signal, 127... Analog comparator, 131... Input binary image, 132... MxN scanning window, 141... Wiring pattern, 151... Thin wiring pattern, 152...・Thick wiring pattern, 153.15
4... Ridge point, 155.156... 5x5 scanning window,
174.175...Ridge point, 173...5x5 scanning window.

Claims (3)

[Claims] (1) An image input means for optically detecting and photoelectrically converting a wiring pattern formed on a printed circuit board, a binarization means for converting a grayscale image from the image input means into a binary image, and a wiring pattern a distance image conversion means for converting each pixel into a distance image indicating a distance value from the background; and a distance image conversion means that detects a ridge point in the distance image and compares the distance value of the ridge point with an arbitrary set threshold value to determine the wiring pattern width. A wiring pattern inspection device comprising a defect detection means for detecting a defect. (2) The wiring pattern inspection apparatus according to claim 1, wherein the defect detection means compares the distance value of the ridge point with a plurality of arbitrary threshold values to detect defects of a plurality of wiring pattern widths. (3) The defect detection means extracts features such as branches and terminations of the ridge point image and detects defects such as short circuits and disconnections in the wiring pattern. Wiring pattern inspection device.