TWI496111B - Bent pin inspection method - Google Patents

Description

Pin bending detection method

The present invention relates to a method for detecting a bending of a stitch, and more particularly to a method for detecting a bending of a stitch applied to a slot of a Central Processing Unit (CPU).

The slots of the current mainstream Central Processing Unit (CPU) can be divided into 1366/1567/2011 pins. Since the number of pins in these slots is very large, pin shifting, foreign matter intrusion, and stitch damage often occur. At present, the method used to detect the pins of the slot is to use manual visual inspection, but this method is not reliable and often causes quality problems in subsequent CPU tests.

In general, the spacing between the pins and the pins is about 40 mils (about 1.016 mm). Among the various adverse conditions such as skewing or shadowing, pressing or upturning, which are often seen in stitches, some offsets are only 4 to 5 mils, which is difficult to detect by the above-mentioned manual visual method.

Furthermore, a single slot contains less than a thousand pins, and more than two thousand pins. If the inspection is performed by manual visual inspection, not only is the time slow, but also a defective product that is misjudged as passing the detection is prone to occur.

In addition, although conventional automatic optical inspection (AOI) equipment is currently available to detect slots, the cost is quite high (more than one million dollars). During the detection process, the conventional automatic optical inspection device must detect another slot before moving to another slot for detection. The more the number of slots to be tested, the longer the detection time required. The problem. Moreover, the fixtures required for conventional automatic optical inspection equipment have a large space.

The present invention provides a method of detecting a stitch bend which is applied to a slot. The slot contains a plurality of pins. The pin bending detection method includes: generating a binarized image about a slot, wherein the binarized image has a stitch matrix pattern; generating a plurality of reference lines based on a contour feature of the stitch matrix pattern; and calculating a complex number on the binarized image based on the reference line The reference point of the group, wherein the reference points of the complex array respectively correspond to the stitches; whether the grayscale value of each set of reference points is 255; and if not, the slot is determined to be seriously unqualified.

In an embodiment of the invention, the step of generating a binarized image about the slot includes: capturing a color image of the slot; converting the color image to a grayscale image; and converting the grayscale image to a binarized image.

In an embodiment of the present invention, the step of converting the grayscale image into a binarized image comprises: dividing the grayscale image into a plurality of image blocks; and defining a plurality of grayscale threshold values corresponding to the image blocks; And binarizing the image block according to the grayscale threshold value, thereby obtaining a binarized image.

In an embodiment of the invention, the range of the gray scale threshold is It is 130~150.

In an embodiment of the invention, the step of generating a plurality of reference lines based on the contour features of the stitch matrix pattern comprises: rotating the binarized image based on the reference line, causing the stitch matrix pattern to be positive.

In an embodiment of the invention, each of the stitches described above includes a needle tip and a needle body. Each set of reference points includes a first reference point and a plurality of second reference points. The first reference point corresponds to the needle tip of the corresponding stitch, and the second reference point corresponds to the needle body of the corresponding stitch.

In an embodiment of the present invention, if the determining result of the step of determining whether the grayscale value of each set of reference points is 255 is yes, the method for detecting the stitching of the stitching further comprises: capturing a plurality of the plurality of reference points based on the first reference point respectively a pixel matrix, wherein each pixel matrix has a tip pattern; determining whether a shape similarity between a tip pattern in each pixel matrix and a tip pattern in any adjacent pixel matrix is greater than a predetermined ratio; and if not, Then it is determined that the slot is slightly unqualified.

In an embodiment of the invention, the predetermined ratio is 70% to 80%.

In an embodiment of the invention, each of the first reference points is located at a center of a corresponding pixel matrix.

In an embodiment of the invention, the contour feature of the stitch matrix pattern is a boundary of the stitch matrix pattern.

In summary, the method for detecting the stitch bending provided by the present invention first performs image capturing and image processing on the slot, thereby obtaining a binarized image of the slot. Next, the stitch bending detection method is performed on the binarized image. Calculate the reference point corresponding to the normal stitch position. Finally, the stitch bending detection method determines whether the grayscale value of the reference point is 255 (ie, white), and if not, it represents that the stitch corresponding to the reference point is offset. Thereby, the stitch bending detecting method of the present invention can detect a small amplitude shift phenomenon (for example, angle skew or shadow, depression or upturn, etc.) which is difficult to observe by manual visual observation, thereby reducing the occurrence of misjudgment loss. Moreover, the stitch bending detecting method of the present invention can adopt a fully automatic optical identification technology, thereby greatly reducing the detection time. Furthermore, the method for detecting the stitch bending of the present invention requires only a low-cost optical jig combination program, and can replace the cost of the construction of the conventional automatic optical detecting device by more than one million yuan.

1‧‧‧ binarized imagery

10‧‧‧ pixel matrix

2‧‧‧ Pin matrix pattern

20‧‧‧ stitch pattern

200‧‧‧ pinpoint pattern

202‧‧‧ needle pattern

22‧‧‧Interval

3‧‧‧ reference line

30‧‧‧First reference point

32‧‧‧second reference point

S100~S122‧‧‧ Process steps

FIG. 1 is a flow chart showing a method of detecting a stitch bending according to an embodiment of the present invention.

Fig. 2 is a detailed flow chart showing the method of detecting the stitch bending in Fig. 1.

Figure 3 is a schematic diagram showing the binarized image of the slot.

Figure 4 is another schematic diagram showing the binarized image in Figure 3, wherein the reference lines are indicated by white lines.

Fig. 5 is a partially enlarged view showing the binarized image in Fig. 3.

Figure 6A is a schematic diagram showing a pixel matrix.

FIG. 6B is a schematic diagram showing another adjacent pixel matrix.

The embodiments of the present invention are disclosed in the following drawings, and for the purpose of illustration However, it should be understood that these practical details are not intended to limit the invention. That is, in some embodiments of the invention, these practical details are not necessary. In addition, some of the conventional structures and elements are shown in the drawings in a simplified schematic manner in order to simplify the drawings.

Please refer to Figure 1, Figure 2 and Figure 3. FIG. 1 is a flow chart showing a method of detecting a stitch bending according to an embodiment of the present invention. Fig. 2 is a detailed flow chart showing the method of detecting the stitch bending in Fig. 1. FIG. 3 is a schematic diagram showing the binarized image 1 of the slot.

As shown in FIG. 1, FIG. 2, and FIG. 3, in the present embodiment, the stitch bending detecting method includes steps S100 to S104.

Step S100: Shoot a color image of the slot.

Step S102: Convert the color image into a grayscale image.

Step S104: Convert the grayscale image into the binarized image 1.

Thereby, a binarized image 1 of the slot can be generated (as shown in FIG. 3). Among them, the binarized image 1 has a stitch pattern 2 . The stitch matrix pattern 2 is composed of a plurality of stitch patterns 20. The gray matrix value of the stitch matrix pattern 2 is 255 (that is, white), and the grayscale value of the region other than the stitch matrix pattern 2 is 0 (that is, black).

In the actual process of generating the above-described binarized image 1, a high-resolution microscope (for example, a resolution of 1280×1024 pixels) can be used, and a special jig, a graphic application interface, and an algorithm are developed to achieve the above object. dedicated The fixture can include a reticle and an auto-dimming device to automatically adjust the light source so that the light source can be strobed along the stitching direction of the slot to obtain a better quality color image.

However, the color images captured in reality are often limited by environmental factors (for example, the number of light sources, the position of the light source, the type of slots, the shape of the stitches, etc.), resulting in uneven distribution of light and dark in various areas of the color image. If the grayscale image is converted into the binarized image 1, the binarization processing of the image is performed only by the fixed grayscale threshold value, and the stitch matrix pattern 2 on the binarized image 1 may be caused. The sizes of the regions vary and cause subsequent misjudgments.

Therefore, in order to solve the above problem, in the present embodiment, the step S104 of the stitch bending detecting method includes steps S104a to S104c as shown in FIG.

Step S104a: dividing the grayscale image into a plurality of image blocks.

Step S104b: Defining a plurality of grayscale threshold values respectively corresponding to image blocks.

Step S104c: Binarize the image blocks according to the grayscale threshold values, and obtain the binarized image 1.

Therefore, according to the stitch bending detection method of the present invention, the darker image block in the grayscale image can be binarized with a relatively low grayscale threshold value, and the relatively bright image in the grayscale image can be made. The block is binarized with a relatively high grayscale threshold. Therefore, when the grayscale image is converted into the binarized image 1, the problem that the size of each region of the stitch matrix pattern 2 on the binarized image 1 is different can be avoided, and the subsequent issue is caused. Misunderstandings.

In practical applications, the gray scale threshold value ranges from 130 to 150, but the invention is not limited thereto.

Please refer to FIG. 4 , which is another schematic diagram of the binarized image 1 in FIG. 3 , wherein the reference line 3 is represented by a white line.

As shown in FIGS. 1 and 4, in the present embodiment, the stitch bending detecting method further includes steps S106 and S108.

Step S106: generating a plurality of reference lines 3 based on the contour features of the stitch matrix pattern 2.

Step S108: Rotating the binarized image 1 based on the reference line 3 causes the stitch matrix pattern 2 to be rotated.

In the present embodiment, the outline feature of the stitch matrix pattern 2 is the boundary of the stitch matrix pattern 2, and therefore the reference line 3 is located at the boundary of the stitch matrix pattern 2 as shown in FIG.

However, the invention is not limited thereto. In another embodiment, the stitch matrix pattern 2 has distinct two gap portions 22, and the contour features of the stitch matrix pattern 2 are the spacing portions 22 in the stitch matrix pattern 2. The position and geometry of the spacing portion 22 vary depending on the type of slot.

Please refer to FIG. 5 , which is a partial enlarged view of the binarized image 1 in FIG. 3 .

As shown in FIGS. 1 and 5, in the present embodiment, the stitch bending detecting method further includes steps S110 to S114.

Step S110: Based on the reference line 3 on the stitch matrix pattern 2 Compute the array reference point.

Step S112: Determine whether the grayscale value of each group of reference points is 255.

In the present embodiment, each stitch includes a needle tip and a needle body. Therefore, each of the stitch patterns 20 includes the tip pattern 200 and the needle pattern 202 (as shown in FIG. 5). Each set of reference points described above includes a first reference point 30 and a plurality of second reference points 32. The first reference point 30 corresponds to the needle tip of the corresponding stitch, and the second reference point 32 corresponds to the needle body of the corresponding stitch. In practical applications, the positions of the first reference point 30 and the second reference point 32 can be derived by the spacing between the reference lines 3 and the standard size of the slot. If the stitching problem does not occur, each first reference point 30 will be located on the corresponding tip pattern 200, and each second reference point 32 will be located on the corresponding needle pattern 202.

In the present embodiment, the number of the second reference points 32 corresponding to each of the needles is 8, but the invention is not limited thereto.

If the result of the determination in the step S112 is NO, the stitch bending detecting method according to the present invention executes the step S114 of determining that the slot is a serious failure.

In other words, if the result of the determination in step S112 is negative, it means that some of the first reference points 30 are not located on the corresponding tip pattern 200, or some of the second reference points 32 are not located on the corresponding needle pattern 202. Therefore, the skew condition of the stitch can be detected. Moreover, when the first reference point 30 is not located on the corresponding tip pattern 200, or the second reference point 32 is not located on the corresponding needle pattern 202, the skew of the stitch is often serious, so The slot is determined to be seriously unqualified.

On the other hand, if the result of the determination in step S112 is YES, each of the first reference points 30 is located on the corresponding tip pattern 200, and each of the second reference points 32 is located on the corresponding needle pattern 202. In other words, it can be concluded that the pins of the slot do not appear to be severely skewed.

Please refer to Figure 6A and Figure 6B. FIG. 6A is a schematic diagram showing the pixel matrix 10. FIG. 6B is a schematic diagram showing another adjacent pixel matrix 10.

As shown in FIG. 1 , FIG. 6A and FIG. 6B , in the present embodiment, if the grayscale value of each set of reference points is 255 (ie, step S112), the determination result is yes. Then, the stitch bending detecting method further includes steps S116 to S122.

Step S116: The plurality of pixel matrices 10 are retrieved based on the first reference point 30, respectively.

In this embodiment, since the position of each of the tip patterns 200 is different from the position of the corresponding first test site, the present embodiment is such that each first reference point 30 is located at the center of the corresponding pixel matrix 10. As the basis for capturing each pixel matrix 10.

Step S118: It is judged whether the shape similarity between the tip pattern 200 in each pixel matrix 10 and the tip pattern 200 in any adjacent pixel matrix 10 is greater than a predetermined ratio.

In the present embodiment, the number of pixels included in each pixel matrix 10 is 2500 (that is, 50×50), but the present invention is not limited thereto and can be elastically changed as needed.

In general, except for the tip of the needle at the boundary of the stitch matrix pattern 2 In addition to the pattern 200, each of the tip patterns 200 will have four adjacent tip patterns 200 (i.e., up, down, left, and right four tip patterns 200).

If the result of the determination in the step S118 is NO, the stitch bending detecting method according to the present invention executes the step S120 of determining that the slot is slightly unacceptable.

In other words, if the result of the determination in the step S118 is negative, it means that although all the stitches of the slot do not have a serious skew condition, some of the stitches have a slight skew condition which is difficult to detect by manual visual observation, so that the slot can be determined to be slight. Not qualified.

If the result of the determination in step S118 is YES, step S122 is performed in accordance with the stitch bending detecting method of the present invention: it is determined that the slot is acceptable.

On the other hand, if the result of the determination in the step S118 is YES, it means that not only the serious skewing condition of all the stitches of the slot does not occur, but also the condition that each stitch is slightly skewed, so that the slot can be judged to be acceptable.

In the present embodiment, the predetermined ratio for determining the shape similarity of the needle tip pattern 200 is 70% to 80%, but the invention is not limited thereto.

From the above detailed description of the specific embodiments of the present invention, it can be clearly seen that the stitch bending detection method provided by the present invention first performs image capturing and image processing on the slot, thereby obtaining a binarized image of the slot. Next, the stitch bending detection method calculates a reference point corresponding to the position of the normal stitch on the binarized image. Finally, the stitch bending detection method determines whether the grayscale value of the reference point is 255 (ie, white), and if not, it represents that the stitch corresponding to the reference point is offset. Thereby, the stitch bending detecting method of the present invention can detect a small amplitude shift phenomenon (for example, angle skew or shadow, depression or upturn, etc.) which is difficult to observe by manual visual observation, thereby reducing The occurrence of a misjudgment is missing. Moreover, the stitch bending detecting method of the present invention can adopt a fully automatic optical identification technology, thereby greatly reducing the detection time. Furthermore, the method for detecting the stitch bending of the present invention requires only a low-cost optical jig combination program, and can replace the cost of the construction of the conventional automatic optical detecting device by more than one million yuan.

Although the present invention has been disclosed in the above embodiments, it is not intended to limit the present invention, and the present invention can be modified and retouched without departing from the spirit and scope of the present invention. The scope is subject to the definition of the scope of the patent application attached.

S100~S122‧‧‧ Process steps

Claims (9)

A pin bending detection method is applied to a slot, the slot includes a plurality of pins, each of the pins includes a tip and a pin, and the pin bending detection method includes: generating a binarization about one of the slots An image, wherein the binarized image has a stitch matrix pattern; a plurality of reference lines are generated based on the contour features of the stitch matrix pattern; and the complex array reference points are calculated on the binarized image based on the reference lines, wherein the groups The reference points respectively correspond to the plurality of pins, each of the group reference points includes a first reference point and a plurality of second reference points, the first reference point corresponding to the corresponding pin point of the pin, and the second reference The point corresponds to the corresponding pin of the pin; whether the grayscale value of each of the group reference points is 255; and if not, the slot is determined to be severely unacceptable. The method for detecting a stitch bending according to the first aspect of the invention, wherein the step of generating the binarized image of the slot comprises: capturing a color image of the slot; converting the color image into a grayscale image; And converting the grayscale image to the binarized image. The method for detecting a stitching according to the second aspect of the invention, wherein the step of converting the grayscale image into the binarized image comprises: dividing the grayscale image into a plurality of image blocks; Defining a plurality of grayscale threshold values respectively corresponding to the image blocks; and binarizing the image blocks according to the grayscale threshold values, respectively, to obtain the binarized image. The method for detecting a stitch bending according to claim 3, wherein the gray scale threshold ranges from 130 to 150. The method for detecting a stitch bending according to claim 1, wherein the step of generating a plurality of reference lines based on the contour feature of the stitch matrix pattern comprises: rotating the binarized image based on the reference lines to cause the stitch matrix pattern Turn positive. The stitch bending detecting method according to claim 1, wherein if the determining result of the step of determining whether the grayscale value of each of the group reference points is 255 is YES, the stitch bending detecting method further comprises: respectively Extracting a plurality of pixel matrices based on the first reference points, wherein each of the pixel matrices has a tip pattern; determining the tip pattern in each of the pixel matrices and any adjacent pixels Whether the shape similarity of the tip pattern in the matrix is greater than a predetermined ratio; and if not, determining that the slot is slightly unacceptable. The method for detecting the stitch bending described in claim 6 of the patent application scope, The predetermined ratio is 70% to 80%. The stitch bending detecting method of claim 6, wherein each of the first reference points is located at a center of the corresponding pixel matrix. The stitch bending detecting method according to claim 1, wherein the contour feature of the stitch matrix pattern is a boundary of the stitch matrix pattern.