JP7423368B2 - Defect detection device and method - Google Patents

Description

The present invention relates to a defect detection device and a defect detection method using images of a substrate.

Patent Document 1 describes an image classification device used for visual inspection of semiconductor substrates and the like. An imaging device captures an image of the inspection target area on the board, and a trained classifier is used to automatically classify defect attributes on the captured image. The system then outputs classification results indicating the type and area of defects.

JP 2017-54239 Publication

In the image classification apparatus disclosed in Patent Document 1, a board that has a defect and should originally be classified as a defective product may be classified as a non-defective product. For example, a trained classifier can detect foreign matter attached to a substrate (hereinafter also referred to as "floating foreign matter") and conditions in which the coating film on the substrate has peeled off and the base is exposed (hereinafter also referred to as "substrate exposure"). ) is difficult to accurately determine. Similarly, it is difficult to accurately distinguish between fibers attached to a substrate (hereinafter also referred to as "fiber foreign matter") and scratches formed on the substrate. For this reason, there was a risk that those with exposed substrate may be mixed in among those classified as having floating foreign matter. Similarly, there is a possibility that items with scratches may be mixed in with items classified as having foreign fiber particles.

Generally, floating foreign matter and fibrous foreign matter can be removed relatively easily, so products classified as having floating foreign matter or fibrous foreign matter are often classified as non-defective products. On the other hand, products classified as having exposed base or scratches are often classified as defective products. Therefore, there is a risk that defective products may be included among those classified as having floating foreign matter or fiber foreign matter and being classified as non-defective products, which is a so-called detection failure. The present invention has been made in view of these circumstances, and an object thereof is to provide a defect detection device and a defect detection method that can improve the accuracy of detecting defective products.

A defect detection device of the present invention for solving the above problems is a defect detection device using an image of a board, and includes a classification section that classifies the attributes of the image using a trained classifier, and The gist of the present invention is to include an image analysis unit that analyzes images belonging to a specific classification among images.

According to the above configuration, by further analyzing the image classified into a specific attribute by the classification unit in the image analysis unit, it is possible to detect defects that were not detected by the classification unit. Compared to a mode in which defects are detected only by the classification section, the accuracy of detecting defects can be improved. Further, high detection accuracy can be obtained without a large number of learning samples and a long learning time to improve detection accuracy in the classification section.

Regarding the defect detection device of the present invention, it is preferable that the above-mentioned specific classification is a classification including at least one of floating foreign matter and fibrous foreign matter. According to this configuration, it is possible to suitably improve the detection accuracy of defects in at least one of base exposure and scratches.

Regarding the defect detection device of the present invention, it is preferable that the image analysis section analyzes at least one of the color tone of the image or the linear objects present in the image. According to this configuration, it becomes possible to efficiently detect base exposure and scratches.

The defect detection method of the present invention for solving the above problems is a defect detection method using an image of a board, which includes a classification step of classifying the attributes of the image using a trained classifier, and a classification step of classifying the attributes of the image using a trained classifier. The method is characterized by comprising an image analysis step of analyzing images belonging to a specific classification among the images.

According to the above configuration, by including the image analysis step of analyzing images classified into specific attributes, it is possible to detect defects that were not detected in the classification step. Compared to a mode in which defects are detected only by the classification step, the accuracy of detecting defects can be improved. In addition, high detection accuracy can be obtained even without a large number of learning samples and a long learning time to improve detection accuracy in the classification step.

Regarding the defect detection method of the present invention, it is preferable that the above-mentioned specific classification is a classification including at least one of floating foreign matter and fibrous foreign matter. According to this configuration, it is possible to suitably improve the detection accuracy of defects in at least one of base exposure and scratches.

In the defect detection method of the present invention, it is preferable that the image analysis step analyzes at least one of the color tone of the image and the linear objects present in the image. According to this configuration, it becomes possible to efficiently detect base exposure and scratches.

According to the defect detection device and defect detection method of the present invention, it is possible to improve the detection accuracy of defective products.

FIG. 2 is a schematic diagram showing the configuration of a defect detection device. A schematic diagram showing the configuration of a computer. FIG. 2 is a block diagram showing the functional configuration of a defect detection device. (a) is a schematic diagram of a substrate with floating foreign matter, and (b) is a schematic diagram of a substrate with an exposed base. (a) is a schematic diagram of a substrate with fiber foreign matter, and (b) is a schematic diagram of a substrate with scratches. Flowchart showing the flow of defect detection.

An embodiment of a defect detection device will be described.
As shown in FIG. 1, the defect detection device 10 includes an imaging device 20 and a computer 30. The imaging device 20 captures an image of the substrate 11 by capturing an image of the inspection target area on the substrate 11, and inputs the obtained image to the computer 30. The computer 30 classifies the attributes of the image input from the imaging device 20 using a learned classifier. Therefore, the computer 30 functions as a classification section.

Further, the computer 30 analyzes images classified as having floating foreign matter or having fibrous foreign matter as a specific classification among the images classified as non-defective by the classification section. Therefore, the computer 30 also functions as an image analysis section. By analyzing the images in the image analysis section, it is possible to detect images that should originally be classified as defective products, such as images with exposed background or scratches, among the images classified as non-defective products by the classification section.

Here, the substrate 11 on which defects are detected by the defect detection device 10 is not particularly limited, but may include a printed circuit board on which a circuit pattern is formed. The printed circuit board may be a board on which electronic components such as semiconductors are mounted, or a board coated with a solder resist.

The imaging device 20 and computer 30, which are components of the defect detection device 10, will be described.
The imaging device 20 will be explained.

As shown in FIG. 1, the imaging device 20 includes a stage 21 on which the substrate 11 is placed, an imaging device 22 that images the substrate 11 on the stage 21, and an image captured by the imaging device 22 that is input to a computer 30. The input device 23 includes an input device 23 for

The stage 21 may be a fixed stage or a movable stage, and is configured to be able to continuously image the plurality of substrates 11 by moving with the plurality of substrates 11 placed thereon. may have been done. The stage 21 may be configured as a belt conveyor or the like to be provided in the substrate 11 manufacturing equipment so as to be able to continuously image a plurality of substrates 11 during the substrate 11 manufacturing process.

The imaging device 22 is not particularly limited, but an optical or electronic camera can be used.
Examples of the input device 23 include, but are not limited to, a cable for wired communication of images converted into electrical signals or optical signals with the computer 30, and a transmitter for wireless communication with a communication unit 39 of the computer 30, which will be described later. machine can be used.

The computer 30 will be explained.
As shown in FIG. 2, the computer 30 has a general computer system configuration including a CPU 31 that performs various calculation processes, a ROM 32 that stores basic programs, and a RAM 33 that stores various information.

The computer 30 includes a fixed disk 34 that stores information, a display 35 that displays various information such as images, and a keyboard 36a and mouse 36b (hereinafter collectively referred to as the "input unit 36") that accept input from the user. have. The computer 30 also transmits signals between a reading device 37 that reads information from a recording medium 38 readable by the computer 30 such as an optical disk, a magnetic disk, and a magneto-optical disk, and other components of the defect detection device 10. It has a communication section 39 for transmitting and receiving.

A program 40 is read in advance from the recording medium 38 into the computer 30 via the reading device 37 and is stored on the fixed disk 34 . Then, the CPU 31 executes arithmetic processing according to the program 40 while using the RAM 33 and the fixed disk 34.

The functional configuration of the defect detection device 10 will be explained.
As shown in FIG. 3, the defect detection device 10 has a classification section 50, a classifier construction section 60, and an image analysis section 70 as functional configurations. The classification unit 50 and the classifier construction unit 60 classify attributes of the image. The image analysis unit 70 analyzes images belonging to a specific classification.

In FIG. 3, the functional configuration realized by the CPU 31, ROM 32, RAM 33, fixed disk 34, etc. of the computer 30 is surrounded by a dashed rectangle with a symbol.
The classification section 50 will be explained.

The classification section 50 includes a target image storage section 51 and a classification control section 52. The classification control section 52 includes a feature amount calculation section and a plurality of classifiers 54. The classification unit 50 classifies the attributes of the image input from the imaging device 20 using neural networks, decision trees, discriminant analysis, etc., and classifies good products and defective products.

The classifier construction unit 60 will be explained.
The classifier construction unit 60 constructs a plurality of classifiers 54 in the classification unit 50. The classifier construction unit 60 includes a teacher data storage unit 61 and a learning unit 62, and a plurality of classifiers 54 are constructed in advance. Details of the functions realized by these configurations will be described later. Note that these functions may be constructed by a dedicated electric circuit, or a dedicated electric circuit may be partially utilized.

The image analysis section 70 will be explained.
The image analysis section 70 uses the image processing software in the computer 30 to process the images classified as having floating foreign matter or having fiber foreign matter as a specific classification among the images classified as non-defective by the classification section 50. To analyze. Image processing includes, for example, converting tints such as color shading to make it easier to extract a specific color, and measuring the shape of a linear object in an image and converting it into a numerical value. Here, the term "linear object" means an elongated shape like a line. The image analysis software may be located in a terminal such as another computer connected to the computer. As the image analysis software, known image analysis software can be used.

The construction mechanism of the classifier 54 will be explained.
The classifier 54 is constructed by assigning values to parameters included in the classifier 54, determining the structure, and the like. When constructing the classifier 54, a large number of images are stored in the teacher data storage section 61 of the classifier construction section 60 as a preliminary preparation. The large number of images include a front surface image, a back surface image, a color image, a black and white image, a non-defective image (hereinafter also referred to as a "reference image"), a defective image, etc. of the substrate 11. Subsequently, the user teaches a plurality of attribute items for the defect indicated by each defect image. Here, the defect attribute items include the color of the defect, the type of defect, the area where the defect exists, and are evaluation items (viewpoints of evaluation) regarding the nature and characteristics of the defect. For example, the defect color has attributes such as a reddish color caused by peeling off the solder resist on the surface of the substrate 11 and exposing the copper wiring. The types of defects include attributes such as misalignment of electronic components on the board 11, chipping of the outer shape, scratches, and exposure of the base. Furthermore, the region where the defect exists has attributes such as the front surface, the back surface, the center portion, and the edge portion.

A plurality of attribute items to be taught (hereinafter referred to as "setting attribute items") are determined by user input via the input unit 36. In teaching about setting attribute items, first, a plurality of defect images are displayed on the computer display. Then, the user performs an input through the input unit 36 indicating one attribute in each setting attribute item for each defect image. That is, input of attributes for each set attribute item for each defect image is accepted by the input unit 36. Hereinafter, the attributes input for the setting attribute items will be referred to as "teaching information." The teaching information regarding multiple setting attribute items for each defective image and the defective image (that is, the defective image to which multiple setting attribute items have been taught, hereinafter referred to as the "teacher image") are mutually exclusive. The data are associated with each other and stored in the teacher data storage unit 61 as a teacher data set 63. In this way, a plurality of teacher data sets 63 are prepared.

In addition, the target image storage unit 51 of the classification unit 50 stores a difference image obtained by comparing the captured image and a reference image, and a mask image obtained by binarizing the difference image as image set data 51a. be remembered. These image set data 51a are also prepared as a teacher data set 63.

The feature amount calculation section of the classification section 50 obtains the feature amounts used in the classifier 54 from the images included in the teacher data set 63. The learning section 62 of the classifier construction section 60 trains the classifier 54 using the teaching information of the defect type in all the teacher data and the feature amounts acquired by the feature amount calculation section. As a result, a plurality of classifiers 54 are individually constructed, each of which performs classification regarding a plurality of setting attribute items.

The image analysis mechanism of the image analysis section 70 will be explained.
FIG. 4(a) shows a schematic diagram of the substrate 11 with floating foreign matter, and FIG. 4(b) shows a schematic diagram of the substrate 11 with exposed underlying material. FIG. 5(a) shows a schematic diagram of the substrate 11 with fiber foreign matter, and FIG. 5(b) shows a schematic diagram of the substrate 11 with scratches.

It is difficult for the classification unit 50 of the computer 30 to accurately distinguish between the floating foreign matter 12 and the exposed base 13. Similarly, it is difficult to accurately distinguish between the fiber foreign matter 14 and the flaw 15. Therefore, there is a risk that an image with an exposed background may be mixed in with an image classified as having a floating foreign object, and an image with a scratch may be mixed in with an image classified as having a fiber foreign object.

In areas where the underlying layer is exposed, for example, the underlying copper wiring is exposed and tends to turn reddish. Therefore, for example, by performing digital image processing to convert the color of the image and making it easier to extract red, it becomes easier to distinguish between the floating foreign matter 12 and the exposed background 13.

Furthermore, the scratches 15 on the substrate 11 tend to be formed in a straight line, whereas the foreign fiber particles 14 tend to have a curled shape with a plurality of bends. Therefore, by measuring the straightness of the linear object in the image using the Hough transform, it becomes easier to distinguish between the fiber foreign object 14 and the flaw 15.

Although the method for measuring straightness is not limited to Hough transform, for example, a plurality of locations in the longitudinal direction of a linear object are displayed as dots, and a virtual line is created by connecting the plurality of dots. Straightness can be evaluated by setting a threshold regarding the straightness value and comparing virtual lines.

The flow of defect detection using the defect detection device will be explained.
A flowchart is shown in FIG.
(Step S101)
Step S101 is a step of capturing an image of the board with an imaging device. The obtained image is input to a computer through an input device.

(Step S102)
Step S102 is a classification step in which images are classified by the classification unit of the computer. Images are classified into non-defective products and defective products, and from among the images classified as non-defective products, images with floating foreign matter and those with fibrous foreign matter are further classified as images belonging to a specific classification.

(Step S103)
Step S103 is an image analysis step in which the images classified as non-defective in step S102 and further classified into those with floating foreign matter and those with fiber foreign matter are analyzed using image analysis software. Using image analysis software in a computer, images with exposed substrate are detected from images classified as having floating foreign matter and classified as defective products. Similarly, images with scratches are detected from the images classified as having fiber foreign matter and classified as defective products.

(Step S104)
Step S104 stores the images classified as defective products in step S102 as defective products, and stores the images classified as non-defective products in the computer as non-defective products. It is a step.

Further, in step S103, images classified into those with exposed background and those with scratches are classified as defective products, and images which are not classified into those with exposed base and scratches are stored in the computer as non-defective products.

Through the above flow, defects can be detected. Furthermore, in step S104, instead of storing the classification results in the computer, the classification results may be displayed on the computer display or output to another device.

The functions and effects of this embodiment will be described.
(1) By performing image analysis in the image analysis unit of the computer on images classified as having floating foreign matter by the classification unit of the computer, it is possible to detect images with exposed background. Similarly, by performing image analysis in the image analysis section on an image classified by the classification section as having a fiber foreign object, it is possible to detect an image with a flaw. Since it is possible to detect defects that were not detected by the classification section, the accuracy of defect detection can be improved compared to a mode in which defects are detected only by the classification section.

(2) In a mode in which defects are detected only by the classification section of a computer, it is necessary to train the classifier using a huge amount of training data in order to improve the defect detection accuracy. Therefore, it takes time and effort to train the classifier. In contrast, the defect detection device of this embodiment uses a trained classifier to classify non-defective products and defective products, and then analyzes images belonging to a specific classification using image analysis software. The detection accuracy can be improved in a shorter time and with less effort.

This embodiment can be modified and implemented as follows. This embodiment and the following modified examples can be implemented in combination with each other within a technically consistent range.
- In the present embodiment, the defect detection device includes an imaging device and a computer, but is not limited to this aspect. The defect detection device may not include an imaging device and may consist only of a computer. For example, an imaging device may be provided separately from the defect detection device, and the image of the board taken by the imaging device may be input to the separate defect detection device to detect defects. .

- In this embodiment, the classification unit, classifier construction unit, and image analysis unit that constitute the defect detection device are provided in one computer, but the present invention is not limited to this aspect. The classification unit, classifier construction unit, and image analysis unit may be provided in different computers.

- In the present embodiment, the specific classifications analyzed by the image analysis unit are both the presence of floating foreign matter and the presence of fibrous foreign matter, but the present invention is not limited to this aspect. The product may be categorized as either having floating foreign matter or having fibrous foreign matter, or may be classified as non-defective products other than having floating foreign matter and other than having fibrous foreign matter.

- Image analysis in the image analysis unit is not limited to the color of the image or the straightness of linear objects.
- In this embodiment, image analysis is performed after classification by a computer classifier, but the invention is not limited to this mode. Before classification is performed by a computer classifier, image analysis may be performed separately to perform classification, and then classification by the computer classifier and analysis by an image analysis unit may be performed.

- In the present embodiment, the classification unit and the image analysis unit perform classification based on images obtained by the same imaging device, but the invention is not limited to this aspect. The classification unit and the image analysis unit may perform classification based on images obtained by different imaging devices. For example, the image analysis section may use an imaging device that can capture images with higher image quality.

- In this embodiment, among the images classified as non-defective by the classification unit, image analysis is performed on images with floating foreign matter and with fiber foreign matter as specific classifications, and defects are detected. but not limited to. Image analysis may be performed on images classified as defective products by the classification unit to detect non-defective products. For example, image analysis may be performed on images classified as having an exposed background and having scratches as specific classifications among images classified as defective by the classification unit. Also in this aspect, it is possible to improve the detection accuracy of defective products.

Examples that further embody the above embodiment will be described below.
(Example 1)
Using the defect detection device of this embodiment, defects in a solder resist substrate on which electronic components were mounted were detected.

First, the board was imaged with an imaging device, and a trained classifier was used to classify good products and defective products. Next, among the images classified as non-defective products, images classified as containing floating foreign matter and containing fibrous foreign matter were analyzed using known image analysis software, and classified into non-defective products and defective products.

For images with floating foreign matter, the color of the image was converted to make it easier to extract red.
For images with fiber foreign matter, Hough transform was performed to display multiple locations in the longitudinal direction of linear objects in the image as dots, and a virtual line connecting the multiple dots was created. The straightness of the linear object was evaluated using a threshold value regarding the straightness value.

(Evaluation test)
Using the defect detection device of this embodiment, the inspectors re-inspected the substrates that were classified as non-defective using a microscope and classified them into non-defective and defective products.

The percentage of items that were classified as defective by the inspector's inspection among the items that were classified as non-defective by the defect detection device was defined as the leakage rate (dpm). Note that (dpm) means the number of defective products included in 1 million non-defective products. In Example 1, the leakage rate was 350 dpm.

(Comparative example 1)
Comparative Example 1 is an example in which non-defective products and defective products are classified using only a trained classifier using the defect detection device of this embodiment. That is, in Comparative Example 1, image analysis using image analysis software was not performed. In Comparative Example 1, the leakage rate was 400 dpm.

DESCRIPTION OF SYMBOLS 10... Defect detection device, 11... Board, 50... Classification part, 54... Classifier 70... Image analysis part.

Claims (4)

A defect detection device based on an image of a board,
a classification unit that classifies attributes of the image using a trained classifier;
Among the images classified by the classification section, images classified as having floating foreign matter are analyzed as images belonging to a specific classification, and images with exposed background are detected and classified as defective products, or identified. What is claimed is: 1. A defect detection device comprising: an image analysis unit that analyzes images classified as having fiber foreign matter as images belonging to the classification, and detects images with scratches and classifies them as defective products . The defect detection device according to claim 1 , wherein the image analysis section analyzes at least one of the color of the image and the linear objects present in the image. A defect detection method using an image of a board,
a classification step of classifying attributes of the image using a trained classifier;
Among the images classified in the classification step, images classified as having floating foreign matter are analyzed as images belonging to a specific classification, and images with exposed background are detected and classified as defective products, or identified. A defect detection method comprising: an image analysis step of analyzing an image classified as having a fiber foreign object as an image belonging to the classification, and detecting an image with a scratch and classifying it as a defective product . 4. The defect detection method according to claim 3 , wherein the image analysis step analyzes at least one of a color tone of the image and a linear object present in the image.