JPH03183147A - Inspection system of ic package - Google Patents

Abstract

PURPOSE:To judge whether a pattern under test is good or defective by a method wherein the number of pattern images of the pattern under test is counted and the number is compared with the number of pattern images of the pattern under test which has been judged to be good. CONSTITUTION:An image data of patterns, under test, of an IC package 1 whose picture has been takes by using a CCD camera 3 is written into an image memory 4. The image memory 4 is an 8-bit memory and has an address composed of (m) lines and (n) rows. When a pattern of the pattern image which has been written in this way is swollen excessively and it is short-circuited with an adjacent pattern, the total number of images of the patterns under test becomes smaller than the number of the patterns which have been judged to be good. Inversely, when one pattern is broken and separated into a plurality of parts, the number of images of the patterns under test becomes larger than the number of images of the patterns which have been judged to be good. Consequently, when the number of picture elements of the patterns is compared with the number of picture elements of the images of the patterns under test which have been judged to be good, it is possible to judge whether the patterns under test are good or defective.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an IC package inspection system.

[Background of the invention] Ceramic packages are widely used to house ICs (integrated circuits).
In order to electrically connect the package to an external circuit, various patterns are printed and plated on the surface of the package. There are some defects in this pattern, such as parts being missing, parts bulging out excessively, the entire pattern not being printed, or not being plated. Therefore, this defect must be inspected, and in recent years it has been considered to use a television camera to inspect this printed pattern.

The present invention aims to more accurately and efficiently inspect IC packages using this television camera.

[Means for Solving the Problem] In order to achieve the above object, in the present invention, the number of pattern images of the pattern to be inspected is counted, and this number and
By comparing the number of pattern images of the pattern to be inspected that has passed, it is determined whether the pattern to be inspected is acceptable or defective. In addition, the number of pixels of the pattern image to be inspected is counted in a specific coordinate direction, and this number of pixels is compared with the number of pixels of the image of the pattern to be inspected that has passed to determine whether the pattern to be inspected is passed or defective. It is something.

[Operation] As a result, for example, if a pattern is excessively expanded and short-circuited with an adjacent pattern, the total number of pattern images to be inspected will be smaller than the number of pattern images that have passed. If one or more patterns are completely dropped,
After all, the total number of pattern images to be inspected is smaller than the number of pattern images that have passed. Conversely, if one pattern is broken and separated into multiple parts, the total number of pattern images to be inspected will be greater than the number of pattern images that have passed. In this way, it can be determined that the product is defective. Also,
If a pattern is partially missing or bulges out, the number of pixels in the pattern image to be inspected may be smaller or larger than the number of pixels in the image of the pattern to be inspected, resulting in a defective product. It can be determined that

[Example] Hereinafter, an example embodying the present invention will be described with reference to the drawings.

FIG. 1 shows an IC package 1.2. IC
The package 1 is formed with a pattern 10 such as a rectangular, rectangular frame, or block shape, and the IC package 2 is formed with a circular,
A ring-shaped, circular block-shaped, etc. pattern 10 is formed. This IC package 1.2 is photographed and inspected from the top, side, etc. using a CCD camera 3.

FIG. 2 shows a circuit diagram of the IC package inspection device. The image data of the pattern to be inspected of the IC package 1.2 photographed by the CCD camera 3 is stored in the image memory 4.
will be written to. This image memory 4 is an 8-bit memory and has addresses in m rows and n columns. The m rows and n columns correspond to the m rows and n columns of the number of photographing pixels of the CCD camera 3.

Data indicating the brightness of each pixel of the photographed pattern to be inspected is written into each address of the image memory 4. This brightness data takes values from 0 to 255. This image memory 4 may have a configuration other than 8 bits. A plurality of image memories 4 are preliminarily provided.

The image data in the image memory 4 is stored in the address counter 5.
After being read out by the CPU 6 and subjected to image data processing, the density distribution memory 7 and the binarized image memory 8
, is written into the label memory 9. address counter 5
has a counter bit of m+n bits or more, and outputs each data of an m-bit row address and an n-bit column address to the image memory 4.

In the density distribution memory 7, as shown in FIG. 3, the number of pixels corresponding to the six brightnesses of the image memory 4 is totaled and stored.

As shown in FIG. 4, the binarized image memory 8 stores, for example, the pattern image
The pattern image 13 is stored in a simplified manner such that three parts are "1 (high level of binary logic level)" and parts other than the pattern image 13 are "0 (low level of binary logic level)". In the label memory 9, as shown in FIG. 5, a number is assigned to each independent pattern image 13, and each pattern image 13 is stored in a distinct manner. Access to the density distribution memory 7, binary image memory 8, and label memory 9 is performed by the CPU 6 through the address counter 5.

The ROMI 1 stores programs for the CPU 6 to perform various processes, and the RAM 12 stores various processing data of the CPU 6. The RAM 12 also stores the number of pattern images 13 to be inspected and the number of pattern images 13 that have passed.
, the number of pixels of the pattern image 13 to be inspected, and the number of pixels of the pattern image 13 of the mask pattern of the accepted product are stored.

Before the inspection process, threshold data T necessary for obtaining a binarized image obtained by simplifying the pattern image 13 is obtained by the following process. Therefore, the CPU 6 first selects the third
The concentration distribution data shown in the figure is created. This is created as follows.

The image data in the image memory 4 is read out for each pixel, and
Aggregation is performed for each brightness level of ~255, and density distribution data is created. This density distribution data is divided into a distribution peak of the relatively bright pattern image 13 and a distribution peak of the image portion other than the relatively dark pattern portion. A small mountain will form between these two mountains. This is a pile of image parts that are defective or poorly printed.

FIG. 6 is a diagram showing a flowchart of IC package inspection processing.

In this process, the CPU 6 expands the pattern image 13 shown in FIG. 7(A) with respect to the image data of the pattern 10 stored in the image memory 4 (step 01). This expansion is a process of expanding the pattern image 13 by one column of pixels toward areas other than the pattern portion. If this process is repeated multiple times, the pattern image 13 will expand by multiple columns. As a result, the pinholes (small holes in the pattern 10) in the pattern image 13 are filled up, as shown in FIG. 7(B).

Next, the pattern image 13 shown in FIG. 7(C) is contracted (step 02). This contraction is caused by pattern image 13
This is a process of shrinking the image by one column of pixels toward areas other than the pattern area. If this process is repeated multiple times, the pattern image 13 will be shrunk by multiple columns. As a result, as shown in FIG. 7(C), the pinholes (small gaps in the pattern 10) in the pattern image 13 are filled, and the size returns to the original pattern image 13 before expansion. Become.

This expansion and contraction process is performed, for example, as follows. The image data of each pixel (t, j) is sequentially read out from the image memory expansion 4, and the surrounding area (i-1, j-1) is
, (i, j-1), (i+1, j-1), (i-1, j
), (i+1, j), (i-1, j+1), (i, j+
1), the image data of (i+1, j+1) are read together, and in the case of expansion, the image data of pixel (iSj) is
(l.

pixel (i, j) in the case of contraction.
The image data of (iS j) is replaced with the minimum value among the image data of (iS j) and the surrounding eight image data and stored in another image memory.

This becomes expanded image data or compressed image data. This expansion and contraction may be performed by another method, or may be performed after conversion to binarized image data, as will be described later.

Then, the original image data in FIG. 7(A) is subtracted for each pixel from the expanded and contracted image data in FIG. 7(C) to extract a grayscale image of the defective part (step 03).
. Next, the grayscale image of the defective part is converted into binary image data using an appropriate threshold value, for example threshold value data T, which will be described later, and each defective part is labeled and distinguished.Step 04) The area of each defective part is determined. It is determined whether or not the value is greater than or equal to the allowable value for an acceptable product (step 05).

This labeling is performed, for example, as follows. The image data of each pixel (t, j) is sequentially read out from the image memory 4 in the vertical direction from the upper left to the lower right. It is determined whether this image data is equal to or greater than an appropriate threshold value, for example, threshold value data T to be described later, that is, whether it is the pattern image 13 or not. If it is equal to or greater than the threshold data T, the label numbers of adjacent pixels are checked.

This label number has already been written in the label memory 9. Adjacent pixels are (i-1, j-1), (l
-1, j), (l-1, j+1), (iSj-1) or (l-1, j), (1%)-1).

If a label number, for example k, has been written to this adjacent pixel, the label number k is written to the address corresponding to the pixel (i, j) in the label memory 9. As a result, the same label number is written for each pixel of one connected pattern image 13. The label numbers of adjacent pixels are all “0”.
”, write a new label number that has not been used before. For example, if label numbers "1" to "3" have been written so far, a new label number "4" is written. As a result, as shown in FIG. 5, a different label number is assigned to each independent pattern image 13 that is not connected.

Furthermore, if a plurality of different values are written in the label numbers of adjacent pixels, the minimum value of the plurality of label numbers is written. Then, the label number of the pattern image 13 on which a label number other than the minimum value has already been written is rewritten to the label number of the minimum value. As a result, the pattern image 13 to which a specific label number was initially assigned becomes
If it is later determined that they are connected, they will be reallocated to the same label number.

Finally, label memory 9 allocated as above
Read out the label numbers of and number them serially starting from "1" starting from the smallest one.

This labeling may be performed based on the binarized image data in the binarized image memory 8 instead of the image data in the image memory 4. In this case, it is not necessary to determine whether or not the threshold value data 77 is greater than or equal to the threshold value data 77.

Moreover, calculation of each area of each pattern image 13 after the above-mentioned labeling is performed as follows.

Each label number is sequentially read out from the label memory 9, and the value of each label number is counted. This total value represents the area of each pattern image 13.

In the inspection process of steps 01 to 05, pattern 10
Pinholes (small holes in the pattern 10) within the area are inspected. This process has the advantage that it is not affected by non-uniform illumination conditions because it is not converted to binary image data until the end.

The process for obtaining the threshold data T is as follows.

First, from the density distribution data, the maximum value Xmaxs and the mode Xmode of the density distribution data of the pattern image 13 shown in FIG. 3 are determined. Next, from the following formula, pattern image 1
The minimum value Xm1n of the concentration distribution data of No. 3 is determined.

min mXmode-aX (Xmax-Xmode) (a is a constant) And between the minimum value Xmin and the maximum value Xmax,
Find the average value X and standard deviation C, and use the following formula to calculate the threshold data T
seek.

T - The data is converted into binarized image data (step 07) and written into the binarized image memory 8 (step 08). Then, as shown in FIG. 8(B), this binary image data is subjected to the above-described expansion process several times (step 09). Next, the above-described labeling process is performed (step 10), and it is determined whether the number of labels matches the number of labels of an acceptable product (step 11). The number of labeled items for this passed product is
It is manually created and memorized in advance. Also, the labeling is 2
The process may be performed on the image data in the image memory 4 instead of the digitized image data.

In the inspection process of steps 06 to 11, pattern 10
It is determined whether the pattern 10 is excessively expanded and short-circuited with the adjacent pattern 10, the pattern 10 falls off, or one pattern 10 is greatly chipped and separated into multiple parts. That is, if the pattern to be inspected 10 is short-circuited, the pattern to be inspected 10.
...The total number of labeled items is less than the number of labeled items of acceptable products. Also, if one or more patterns 10 are completely missing, the pattern 10 to be inspected...
・The total number of labeled items is less than the number of labeled items of acceptable products. On the other hand, if one pattern 10 is broken and separated into a plurality of parts, the number of labeled parts of the pattern to be inspected 10 as a whole will be greater than the number of labeled parts of an acceptable product. In this way, it can be determined that the product is defective.

Note that although the above expansion process may be omitted, it is better not to omit it. Even if the pattern 10 is excessively expanded and does not short-circuit with the adjacent pattern 10, even in the case of a very close short-circuit, the pattern 10 connects with the adjacent pattern 10 by the expansion process and can be detected as a defective product. This is because this near short also does not have good performance as a product.
This is to exclude it from the approved products. Further, the CPU 6 reads the binarized image data from the binarized image memory 8 pixel by pixel (step 12), and totals the number of pixels of the binarized image data of rlJ in the Y-axis direction, that is, in the vertical direction (step 12). 13). As shown in FIG. 9, this totaling is performed by counting the number of binary image data of "1" for a group of pixels having the same X-axis coordinate value, and totaling this for each X-coordinate.

Then, this number of pixels in the Y-axis direction is compared with the number of pixels in the Y-axis direction of the mask pattern of the accepted product, and it is determined whether there is a deviation of more than a certain value (step 14).

This deviation is determined based on whether it exceeds the number of pixels of an acceptable product. Of course, the difference may be determined based on whether the difference is less than a certain value or a certain ratio with respect to the number of pixels of a passing product. In this determination of the number of pixels, in order to correct the deviation between the pixel number distribution of the pattern to be inspected 10 and the pixel number distribution of the mask pattern of the accepted product, a deviation correction in the X-axis direction, which will be described later, may be performed. In addition, the number of pixels of the mask pattern of the passed product is inputted and stored in advance, and the number of pixels of the mask pattern of the passed product is
The minimum value, average value, mode, etc. of the number of pixels of 3 are used.

In the inspection process of steps 12 to 14, pattern 10
Large chips, drops, large breaks, and large shorts are identified. In addition, the entire pattern 10 to be inspected has short-circuits or omissions of the pattern 10, and 1
If one pattern 10 is largely chipped and separated into multiple parts, the total number of labeled patterns 10 to be inspected will coincidentally match the number of labeled parts of the passed product, and steps 09 to 11 will be performed. Although the product is determined to be an acceptable product, it can be determined as a defective product in steps 12 to 14.

Note that an expansion process may be performed in steps 12 to 14, or the image data of each pixel (i, j) may be sequentially read out from the image memory 4, and whether this image data is equal to or greater than an appropriate threshold value, for example, the threshold data T described above. No, that is, pattern image 1
It may be determined whether the number of pixels is 3 or not, and the number of pixels may be totaled in the Y-axis direction, that is, in the vertical direction.

FIG. 10 shows another embodiment. This inspection process is performed as follows.

The above-described expansion process is performed n times on the pattern image 13 of the frame-shaped pattern to be inspected 10 in the binarized image memory 8 (
(FIG. 10(B)), the pattern image 13 of the mask pattern of the accepted product stored in advance is superimposed (first
Figure 0 (C)). This superposition is performed by performing a logical sum on each pixel of each binary image data of the pattern image 13 of the pattern to be inspected 10 and each binary image data of the pattern image 13 of the mask pattern of the accepted product. It is done by

After this, the above-mentioned contraction process is performed n+1 times (Fig. 10 (
D)), this contracted image data and the original image data are r
lJ rOJ Performs a logical product with the inverted image data. As a result, only the defective parts are extracted (Fig. 10 (
E)). When determining whether the area of each defective part is equal to or larger than the allowable value for a passing product, the pattern image 13 of the mask pattern of the passing product is added to the expanded image of the pattern image 13 of the pattern to be inspected 10.
The reason for superimposing them is to be able to cope with even if there are variations in the size of each pattern to be inspected 10 and a slight positional shift.

Further, when the expanded image of the pattern image 13 of the pattern to be inspected 10 and the pattern image 13 of the mask pattern of the accepted product are superimposed, a process for correcting positional deviation is performed. This deviation correction amount ΔX1ΔY is determined by the following equation.

ΔX = ((x1+x2)-(X1+X2))/2ΔY = ((yl,+y2)-(Y1+Y2)l/2
Here, xl, x2, yl, y2 are coordinate values outside the frame-shaped pattern to be inspected 10, and Xl, X2, Yl, Y
2 is a coordinate value outside the frame-shaped acceptance pattern 10.

Furthermore, the pattern pixels 13 of the mask pattern of the passed product are as follows:
A logical product of pattern images 13 of a plurality of acceptable products, an average pattern image, and a most frequent pattern image are used.

The present invention is not limited to the above embodiments, and various changes can be made without departing from the spirit of the present invention.

[Effects of the Invention] As described in detail above, according to the present invention, the number of pattern images of the pattern to be inspected is counted, and this number is compared with the number of pattern images of the pattern to be inspected that have passed. It is now possible to determine whether the inspection pattern passes or fails. As a result, for example, if a pattern is excessively expanded and short-circuited with an adjacent pattern, the total number of pattern images to be inspected will be smaller than the number of pattern images that have passed. If one or more patterns are completely dropped, the total number of pattern images to be inspected will still be smaller than the number of pattern images that have passed. Conversely, if one pattern is broken and separated into multiple parts, the total number of pattern images to be inspected will be greater than the number of pattern images that have passed.

In this way, it can be determined that the product is defective. In addition, the number of pixels of the pattern image to be inspected is counted in a specific coordinate direction, and this number of pixels is compared with the number of pixels of the image of the pattern to be inspected that has passed to determine whether the pattern to be inspected is passed or defective. . As a result, if a part of the pattern is missing or bulges out, the number of pixels in the pattern image to be inspected may be smaller or larger than the number of pixels in the pattern image to be inspected. It can be determined that the product is defective.

In this way, IC packages can be inspected more accurately and efficiently.

[Brief explanation of drawings]

1 to 10 show embodiments of the present invention. FIG. 1 is a perspective view of an IC package, FIG. 2 is an overall circuit diagram of an inspection device, and FIG. 3 is a pattern image 13. Fig. 4 is a graph showing the brightness data of each pixel of 2.
5 is a diagram showing the binarized image data of the pattern 10 stored in the digitized image memory 8, FIG. 5 is a diagram showing the label number of each pattern 10 stored in the label memory 9, and FIG. It is a flowchart of the inspection process for the patterns 10... of the IC package, and FIGS. 7 to 10 are diagrams showing the process of inspecting the patterns 10... 1.2...IC package, 3...CCD camera,
4... Image memory, 5... Address counter, 6.
...CPU, 7...Density distribution memory, 8...Binarized image memory, 9...Label memory, 10...Pattern, 13...Pattern image.

Claims (2)

[Claims] (1) A photographing means for photographing the pattern to be inspected; a means for distinguishing between images of the pattern portion and images other than the pattern portion for the pattern to be inspected photographed by this photographing means; a counting means for counting the number of each pattern image; a comparison means for comparing the number of pattern images counted by the counting means with the number of pattern images of the pattern to be inspected that has passed; An IC package inspection system comprising: a determining means for determining whether a pattern to be inspected is acceptable or defective; (2) A photographing means for photographing the pattern to be inspected; and a means for distinguishing between images of the pattern portion and images other than the pattern portion for the pattern to be inspected photographed by the photographing means; a counting means for counting the number of pixels of the pattern image to be inspected in a specific coordinate direction; and a comparison means for comparing the number of pixels of the pattern image to be inspected counted by the counting means with the number of pixels of the pattern image to be inspected that has passed. , and a determining means for determining whether the pattern to be inspected is acceptable or defective according to the comparison result of the comparing means.