JP2002277409A - Inspection device for pattern of printed board - Google Patents

Abstract

PROBLEM TO BE SOLVED: To inspect the wiring pattern of a printed board precisely and accurately at a high speed. SOLUTION: The point on the contour of an object to be inspected in an image is tracked by a contour tracking part 16 to form the coordinates list of the point on the contour. Predetermined feature quantities arithmetically led out using three points in total, that is, the arbitrary one point in the coordinates list and two points separated from one point by a distance(s) before and behind in the coordinates list, are calculated with respect to all of the points in the coordinates list by a feature quantity calculater 18. A flaw structural region is detected and classified on the basis of the calculated feature quantities by a position/class converter 24 for an inspection pattern. The flaw structural region is classified to a class same or near to a class wherein a good article preliminarily obtained with respect to a good article sample by the same processing and stored in a position/class memory 26 for a reference pattern as a position/class list 26A is classified. The region is determined to be the good article by a position/class comparing part 28.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a printed circuit board pattern inspection apparatus for inspecting a wiring pattern on an industrial printed circuit board, and more particularly, to a two-dimensional inspection apparatus which involves deformation of the contour shape of the wiring pattern. The present invention relates to the detection of chipped defects, short-circuits, disconnections, and defects equivalently observed on an image.

[0002]

2. Description of the Related Art As a typical defect of a wiring pattern on a printed circuit board, a short circuit as shown in FIG. 6A, a disconnection as shown in FIG. 6B, and a defect as shown in FIG. There is a lack, a projection as shown in FIG. 3D, a pinhole as shown in FIG.

As a technique for automatically inspecting for such a defect, for example, Ando "Printed board inspection using radial matching", Journal of the Japan Society of Precision Engineering, Vol. 56, No. 8, No. 30
Pages 33 to 33 and a method using a radial matching method as disclosed in Japanese Patent No. 2516392 are known.

In this radial matching method, a plurality of sensors arranged radially are used to detect the center of a pattern, and from this center, three pieces of information of a pattern measurement value, a direction, and a center condition are extracted, and a code is extracted. The inspection is performed by comparing this code obtained from the pattern to be inspected with a code of a dictionary created from a non-defective printed circuit board for the same pattern.

[0005]

However, the above-described radial matching method has a problem that it is not suitable for high resolution because the amount of calculation is large and a large amount of memory is required for storing codes.

Also, different pattern widths are mixed.
It is difficult to handle complicated pattern shapes.

The present invention has been made in view of the above points, and has as its object to provide a printed circuit board pattern inspection apparatus capable of inspecting a wiring pattern on a printed circuit board at high speed, accurately, and accurately.

[0008]

In order to achieve the above object, an apparatus for inspecting a printed circuit board pattern according to the first aspect of the present invention traces a point on an outline of an object to be inspected in an image. Contour tracing means for generating a coordinate list of points on the contour, an arbitrary point in the coordinate list generated by the contour tracing means, and a point separated from the arbitrary one point by s before and after in the coordinate list A characteristic amount calculating unit that calculates a predetermined characteristic amount arithmetically derived using two points in total for all the points in the coordinate list; and the characteristic amount calculated by the characteristic amount calculating unit. And a defect classification means for detecting a defect structure portion and classifying the defect structure based on the defect classification.

In other words, according to the printed board pattern inspection apparatus of the present invention, since the inspection is performed using only the pixels on the contour of the object, complicated processing is performed on all the pixels in the image. A high-speed and high-precision pattern inspection can be performed without performing.

According to a second aspect of the present invention, in the inspecting apparatus for a printed circuit board pattern according to the first aspect, the characteristic amount is a triangular shape having the three points as vertices. the area to generate the area list obtained for all contour points, the index area list on the horizontal axis, when you create a graph of the area value transition taking the area value on the vertical axis, the area value exceeds the threshold value T _v and the difference D of the index falls below the index and then T _v, the integral value and the C portion area value exceeds the threshold T _v, a, the defect classification means, the area value is a predetermined threshold value T _v
Is considered as a defective structure site, and classification is performed using the difference D and the integral value C.

In other words, according to the printed circuit board pattern inspection apparatus of the present invention, since the defect structure is found by using the threshold value _Tv , the classification is performed. Classification can be performed efficiently without doing.

According to a third aspect of the present invention, in the inspection apparatus for a printed circuit board pattern according to the first aspect of the present invention, the inspection apparatus is classified into a class which is the same as or close to a class in which a non-defective product is classified. It is characterized in that a determination means for determining a defective structure portion as a non-defective product is further provided.

That is, according to the printed circuit board pattern inspection apparatus of the present invention, the defect judgment is performed using the class, so that the judgment processing is simplified.

According to a fourth aspect of the present invention, there is provided a printed circuit board pattern inspection apparatus according to the third aspect, wherein the determining means determines a line profile near a portion determined to be defective. In which the line width is measured, and more strict defect determination is performed.

That is, according to the printed circuit board pattern inspection apparatus of the present invention, since the measurement using the line profile is performed only around the part determined to be suspicious, the range to which the line profile is applied is limited. Processing can be performed at high speed.

[0016]

Embodiments of the present invention will be described below with reference to the drawings.

[First Embodiment] FIG. 1A is a block diagram of a printed circuit board pattern inspection apparatus according to a first embodiment of the present invention.

That is, the inspection apparatus for a printed circuit board pattern according to the present embodiment comprises a camera 10, an image memory 12,
Digitizing unit 14, contour tracking unit 16, feature amount calculating unit 18, s value setting unit 20, s value correcting unit 22, inspection pattern position / class conversion unit 24, reference pattern position / class storage unit 2.
6, a position / class comparison unit 28 and a determination unit 30.

Here, the camera 10 includes an image pickup device such as a CCD and shoots an inspection object. The image memory 12 stores an image captured by the camera 10.
The binarizing unit 14 binarizes the image stored in the image memory 12 using a predetermined threshold. in this case,
The above-described method of determining the predetermined threshold value may be performed by, for example, selecting a statistically optimum value by photographing a plurality of inspection targets in advance, or determining the value while the operator manually views the binarization result each time. May be something. By this binarization processing, the result 2
The object to be inspected in the value image has the foreground color, and the other portions have the background color.

The contour tracking section 16 tracks the contour of the inspection target from the image data binarized by the binarizing section 14. That is, as shown in FIGS.
The contour tracking is performed for each object in the value image, the coordinates of the points (contour points) on the contour are obtained, and a contour point sequence coordinate list 16A as shown in FIG. It is stored in a memory (not shown) formed inside the contour tracking unit 16.
FIG. 2B is an enlarged view of the pixels constituting the contour image in the frame surrounded by the dotted line in FIG. 2A, and arrows indicate the order of contour tracking. Here, a general 8-neighbor search is used for the contour tracking method. In FIG. 2C, L is the number of contour points for a certain object.

The feature value calculating unit 18 calculates a feature value, for example, a triangular area, according to the s value as a parameter set by the s value setting unit 20. That is, as shown in FIG. 2D, in the contour point sequence coordinate list 16A generated and stored by the contour tracking unit 16, attention is paid to a certain point P _i (coordinate values (x _i , y _i )). Then, the coordinate values ((x _is , y _is ) and (x _is ) of points P _is and P _{i + s} separated from the point by s before and after in the contour point sequence coordinate list 16A.
_{i +} s, obtaining the triangle _{P i} P is P _{i + s} of the area a _i with y _{i +} s)). The following equation (1) is used for calculating the area a _i .

A _i = (１ ／) {(x _is −x _i ) (y _{i + s} −y _i ) − (y _is −y _i ) (x _{i + s} −x _i )} (1) , in the case of n <0 or n ≧ L for any integer n, it is defined as _{p n = P n mod L (} 2). Here, “n mod L” represents a remainder when n is divided by L.

The area value defined here depends on whether the contour shape is convex inside or outside the object as shown in FIGS. 3A and 3B, respectively. Takes a positive or negative value.

Then, the area of the triangle is obtained for all the contour points, and a triangle area list 18A as shown in FIG. 2E is generated. Store in memory.

Further, as shown in FIG. 3C, the feature quantity calculating section 18 creates a transition graph of the area value using the index of the area list 18A on the horizontal axis and the area value on the vertical axis. I do. Then, it provided a predetermined threshold T _v the area value, a portion that exceeds this threshold T _v are those regarded as "chipping structure". As shown in the figure, two or more chipped structures may exist on the contour of one object.

Then, the following three feature values as shown in FIG. 4A are obtained for each of the detected missing structures, and are stored in a memory (not shown) formed inside the feature value calculation unit 18.

That is, the first feature value is that the area value is equal to the threshold value T.
and index beyond the _v, which is the difference D of the subsequent index falls below the threshold value T _v.

The characteristic amount of the second is the integral value of the portion area value exceeds the threshold value T _v (area) C.

[0029] Then, the feature amount of the third uses an index i _c of the center of the index area value falls below the index and then the threshold T _v exceeding the threshold T _v, see contour point sequence coordinate list This is the point _Pic obtained as a result. In this case, the point _Pic, which is the third feature value, may be stored after returning to the XY coordinate table using the outline point sequence coordinate list 16A.

Here, the s-value setting section 20 is for the operator or the like to set a value according to the size of the inspection object and the size of the chipped defect to be detected. When the magnification of the lens 10A is changed, the size of the object in the obtained image also changes.
The correction of the s value linked to the lens magnification is performed by 2,
The information is supplied to the feature amount calculation unit 18.

Inspection pattern position / class converter 24
Is to classify the missing structure into classes using the feature amount calculated by the feature amount calculation unit 18. That is, as shown in FIG. 4 (B), the value of the feature amount C is the integral value of the portion area value exceeds the threshold T _v, the area value is the threshold value T
_v depending on the value of the feature amount D is the difference of the index falls below the index and then the threshold T _v beyond the,
Mapping is performed on two-dimensional coordinates, and the detected missing structures are classified.

In FIG. 4B, it is assumed that the classes are classified into five stages, that the class 1 has the smallest structure and the class 5 has the most severe degree. If the number of classes is further increased, more detailed classification can be performed, and accordingly, the judgment performance at the time of inspection also improves.

Reference pattern position / class storage section 26
At the time of non-defective learning, the above structure is used to extract the missing structure and its feature amount using the non-defective image obtained by photographing the non-defective sample, and determine the position of the defective structure that should be originally on the pattern contour and its class Then, the position and class of each missing structure in the non-defective sample are received from the inspection pattern position / class converter 24, and a position / class list 26A is generated and recorded as shown in FIG. 1B. It is something to keep. In this case, since only the position and class of each missing structure are recorded, even if the inspection image size becomes very large, the storage capacity is small.

In this non-defective product learning, only one non-defective sample of a printed circuit board may be used, or the same operation may be performed on a plurality of non-defective samples to learn the average value. You may do it.

In this way, after learning the non-defective product, a chipped structure and its feature amount are similarly extracted from the actual printed circuit board to be inspected by using an inspection image obtained by photographing, and the chipped structure which should be originally present on the pattern contour is obtained. The position and its class will be determined.

The position / class comparing section 28 receives the position and class of each of the missing structures on the inspected printed circuit board from the inspection pattern position / class converting section 24, and temporarily detects all the detected missing structures. , Included in the defect set.

The position obtained by the non-defective learning stored in the reference pattern position / class storage unit 26
With reference to the class list 26A, it is determined whether or not the following conditions are satisfied for all the missing structures in the defect set, and the missing structures that satisfy the conditions are sequentially removed from the defect set.

That is, when the position of the chipped structure of interest is P and the class is K, the position P
The distance T _d is recorded that there is lack structure positioned within, and if the class is equal to or less than K-n ₁ or K + n ₂ from excluding the missing structure from the defect accumulating. That is,
A chip defect that is close in position and close in class to that of a good sample is not actually a defect.

Note that n ₁ and n ₂ are natural numbers and can be arbitrarily set by the operator. With this value, the inspection level (severity) can be determined.

Then, the determination unit 30 determines the position of the chipped structure remaining in the defect set as a defect position and outputs (reports) it.

The defects detected here are, in addition to the chipped defect, a short defect as shown in FIG. 4C and a partial defect as shown in FIG. is there. Although these are not chipped structures themselves, they have a local chipped structure at the joint between the defective portion and the wiring pattern, and this is detected.

Of course, the same processing can be performed not only in the case where the area of the missing structure, ie, the triangle, is positive, but also in the case where the area of the convex structure, ie, the triangle, is negative. By performing such a process of detecting a convex structure, it is possible to detect a disconnection defect that is difficult when only the missing structure detection is used, as shown in FIG. The accuracy of detecting a convex defect as shown in FIG.

Furthermore, if the above-described detection of the chipped structure and the detection of the convex structure are used in combination, the detection performance is improved by the complementary effect.

The method of the first embodiment can also be used for a defect whose contour is closed inside the object. That is, as shown in FIG. 5A, a contour that does not exist at the time of learning the good product is generated.
If a chipped structure is detected, it can be immediately determined as a defect.

Thus, a pinhole defect can be detected.

In the first embodiment, an area list is calculated using a contour point sequence list, and the calculated list is used as a basic feature amount, and a feature amount with a higher degree of abstraction is obtained. However, other than such a triangular area, basic feature amounts as shown in FIGS. 5B and 5C can be considered. That is, (B) in the figure shows the angle formed by the three points, and (C) in the figure shows the distance between the straight line formed by the two points and another point as the basic feature amount.

The same detection result can be obtained by using these, but the amount of calculation increases.

[Second Embodiment] Next, a second embodiment of the present invention will be described.
An embodiment will be described.

According to the second embodiment, the line width is measured by a line profile in the vicinity of a portion output as a defect from the determination unit 30 by the processing of the first embodiment, and the defect is more strictly determined. The judgment is performed.

That is, first, a missing structural defect is detected in the same manner as in the first embodiment.

Next, as shown in FIG. 5D, one direction is detected in a direction orthogonal to the detected wiring pattern near the chipped structure.
More than one observation line is set up, and a line profile of luminance along those observation lines is created.

Thereafter, the length of the line profile exceeding a predetermined luminance value is detected as the line width of the wiring pattern.

Then, it is determined whether or not the line width is within an appropriate range, and if the line width is within the appropriate range, it is not reported as a defect.
Report the missing structure as a defect only if it is not in the proper range.

As described above, first, the inspection level is strictly set, and a defect which is considered to be defective is roughly extracted by the method of the first embodiment. By measuring the line width and performing more strict defect determination, the range to which the line profile is applied can be limited, so that the processing can be performed at high speed.

Although the present invention has been described based on the above embodiments, the present invention is not limited to the above embodiments, and various modifications and applications are possible within the scope of the present invention. Of course.

[0056]

As described in detail above, according to the present invention,
It is possible to provide a printed circuit board pattern inspection apparatus capable of inspecting a wiring pattern on a printed circuit board at high speed, precisely, and accurately.

That is, since the inspection is performed using only the pixels on the contour of the object, high-speed and high-precision pattern inspection can be performed without performing complicated processing for all the pixels in the image. Therefore, the processing time can be reduced, and the cost of the apparatus can be reduced.

[Brief description of the drawings]

FIG. 1A is a block diagram of a printed circuit board pattern inspection apparatus according to a first embodiment of the present invention;
(B) is a diagram showing a position / class list.

FIGS. 2A and 2B are diagrams for explaining contour tracing, respectively, and FIG. 2C is a diagram showing a contour point sequence coordinate list;
(D) is a diagram for describing a triangle having a contour point as a vertex as an example of a feature amount, and (E) is a diagram illustrating a list of areas of the triangle.

3 (A) and 3 (B) are diagrams respectively showing a contour shape and a transition of an area feature amount in correspondence with each other, and FIG. 3 (C) is a diagram showing a transition graph of a contour shape and an area value and a missing structure. FIG.

4A is a diagram for explaining a feature amount of a missing structure, FIG. 4B is a diagram for explaining classification of the missing structure, FIG. 4C is a diagram for explaining detection of a short circuit,
(D) is a diagram for explaining the detection of the protrusion, (E) is a diagram for explaining the detection of the disconnection, (F) is a diagram for explaining the detection of the protrusion.

FIG. 5A is a view for explaining a defect inside an object,
(B) is a diagram for explaining an angle formed by three points as basic features other than the area of the triangle, and (C) is a straight line formed by two points as the basic features other than the area of the triangle and another It is a figure for explaining distance to one point, and (D) is a figure for explaining a detecting method of a chipped structural defect in a printed circuit board pattern inspection device concerning a second embodiment of the present invention. .

FIGS. 6A and 6B are diagrams showing typical defects of a wiring pattern on a printed circuit board. FIG.
(B) is a diagram showing a disconnection, (C) is a chip, (D) is a protrusion, and (E) is a diagram showing a pinhole.

[Explanation of symbols]

 Reference Signs List 10 camera 10A lens 12 image memory 14 binarization unit 16 contour tracking unit 16A contour point sequence coordinate list 18 feature amount calculation unit 18A area list 20 s value setting unit 22 s value correction unit 24 inspection pattern position / class conversion unit 26 Reference pattern position / class storage unit 26A position / class list 28 position / class comparison unit 30 judgment unit

 ────────────────────────────────────────────────── ─── Continued on the front page (72) Inventor Yamato Kanda 2-43-2 Hatagaya, Shibuya-ku, Tokyo F-term in Olympus Optical Co., Ltd. (reference) 2F065 AA23 AA49 AA56 BB02 CC01 DD03 DD06 FF04 FF42 JJ03 JJ26 QQ04 QQ21 QQ24 QQ32 SS04 2G051 AA65 AB07 CA03 EA11 EA14 EB01 EC01 ED11 ED22 5B057 AA03 BA11 CA08 CA16 CB06 CB16 CE12 CH11 DA03 DB08 DC04 DC16 DC36

Claims (4)

[Claims] 1. A contour tracing means for tracing a point on a contour of an object to be inspected in an image and generating a coordinate list of points on the contour, and an arbitrary point in the coordinate list generated by the contour tracing means And a predetermined feature amount that is arithmetically derived from the arbitrary point by using a total of three points separated by s before and after in the coordinate list from the arbitrary one point is calculated by using all points in the coordinate list. And a defect classifying means for detecting a defect structural part and classifying the defect based on the feature quantity calculated by the feature quantity calculating means. Board pattern inspection equipment. 2. The feature quantity is obtained by calculating an area of a triangle having the three points as vertices for all contour points, generating an area list, and taking an index of the area list on the horizontal axis and an area value on the vertical axis. when you create a graph of the area value transition, the difference D between the index falls below the index and then T _v the area value exceeds the threshold T _v, the integrated value C of the part area value exceeds the threshold value T _v The defect classifying means determines that the area value is equal to a predetermined threshold value T.
The printed circuit board pattern inspection apparatus according to claim 1, wherein a part exceeding _v is regarded as a defective structure part, and classification is performed using the difference D and the integral value C. 3. The printed circuit board pattern inspection according to claim 1, further comprising determining means for determining a defective structure portion classified as the same or close to the class in which the non-defective product is classified as a non-defective product. apparatus. 4. The inspection of a printed circuit board pattern according to claim 3, wherein said determination means measures a line width in the vicinity of a part determined as a defect by a line profile and performs a more strict defect determination. apparatus.