JP2006078285A - Substrate-inspecting apparatus and parameter-setting method and parameter-setting apparatus of the same - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a technology for automatically generating parameters used for a substrate-inspecting apparatus. <P>SOLUTION: A parameter-setting apparatus considers colors of pixels in a target image as target points, considers the colors of pixels in an excluded image as exclusion points, and maps them to a color space (a color histogram). A color range for dividing the color space and maximizing a difference (a sum of degrees) between the number of the target points and the number of the exclusion points included in them, is found. The found color range is set as a color condition (a color parameter) used for a substrate inspection. Accordingly, the color condition which is one of the inspection parameters is automatically generated. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a technique for automatically generating parameters used in a substrate inspection apparatus.

  Conventionally, there has been proposed a board inspection apparatus for inspecting the mounting quality of a printed circuit board on which a large number of electronic components are mounted. In this type of printed circuit board, “the shape of the solder pile when the electrode part of the electronic component and the land are soldered” is called a solder fillet. However, depending on the wetting of the electrode part of the electronic component, a solder fillet is formed. In some cases, the electronic component and the solder fillet are not in contact with each other. Therefore, in order to inspect the soldering quality, it is necessary to accurately grasp the shape of the solder fillet made of a free curve.

  However, in the conventional board inspection apparatus, since monochrome (single color) single illumination is used as a light source, it is difficult to perform image analysis of the three-dimensional shape of the solder fillet. Therefore, the quality of soldering cannot be determined, and the board inspection apparatus cannot be used practically.

  In order to solve such a problem, the present applicant has proposed a substrate inspection apparatus of the type shown in FIG. 23 (see Patent Document 1). This method is called the three-color light source color highlight method (or simply the color highlight method), and is a technology that obtains the three-dimensional shape of the solder fillet as a pseudo-color image by illuminating the inspection object with light sources of multiple colors. is there.

  It is said that the practical use of automatic inspection of printed circuit boards is practically after the advent of this color highlighting technology. In particular, at the present time when electronic components are miniaturized, it is difficult to visually determine the solder fillet shape, and it can be said that board inspection cannot be realized without a color highlight type substrate inspection apparatus.

  As shown in FIG. 23, the color highlight type substrate inspection apparatus images a light projecting unit 105 that irradiates the inspection object 107 on the substrate 110 with the three primary color lights at different incident angles, and reflected light from the inspection object 107. An imaging unit 106. The light projecting unit 105 includes three annular light sources 111, 112, and 113 that have different diameters and that simultaneously emit red light, green light, and blue light based on a control signal from the control processing unit. Yes. The light sources 111, 112, and 113 are arranged in directions corresponding to different elevation angles when viewed from the inspection target 107 while being centered on the position directly above the inspection target 107.

  When the inspection target (solder fillet) 107 is irradiated by the light projecting unit 105 having such a configuration, light of a color corresponding to the inclination of the surface of the inspection target 107 enters the imaging unit 106. Therefore, as shown in FIG. 24, when the soldering of the electronic component is good / when the component is missing / when the solder is insufficient, the color pattern of the captured image depends on the shape of the solder fillet. A clear difference appears. This facilitates image analysis of the three-dimensional shape of the solder fillet, and makes it possible to accurately determine the presence / absence of electronic components and the quality of soldering.

In the color highlight type substrate inspection system, color parameters (color conditions) representing “the color of a good product that should be” and “the color of a defective product that should be” are set in advance, and the corresponding color parameter is selected from the inspection image. The color area to be extracted is extracted, and the quality is determined based on various feature amounts (for example, area and length) of the extracted area. Therefore, prior to the actual inspection, it is necessary to set the color parameters used for the inspection, the types of feature amounts, the determination conditions (for example, threshold values) for separating good products from defective products, and the like. The color parameters, feature amounts, and determination conditions are collectively referred to as inspection logic or inspection parameters, and setting and adjusting the inspection logic is generally referred to as teaching.

  In order to improve the inspection accuracy, it is important to set the color parameter so that a significant and clear difference appears between the feature quantity indicated by the non-defective product and the feature quantity indicated by the defective product. That is, it can be said that the quality of teaching of color parameters directly affects the inspection accuracy.

Therefore, the present applicant has proposed a tool for supporting the setting of color parameters in the color highlight method as shown in FIG. 25 (see Patent Document 2). In this tool, it is possible to set the upper limit value and the lower limit value of each of the hue ratios ROP, GOP, BOP and lightness data BRT of red, green, and blue as color parameters. The input screen of FIG. 25 is provided with a color parameter setting unit 127 for inputting color parameter setting values, and a setting range display unit 128 for displaying a color range extracted by each set color parameter. ing. The setting range display section 128 displays a hue diagram 134 showing all colors obtained under a predetermined brightness. When the operator sets the upper limit value and lower limit value of each color parameter, Displays a confirmation area 135 surrounding the color extracted by the set color parameter. When the binarization display button 129 is pressed, the extraction result based on the current color parameter is displayed as a binary image. According to this tool, the operator can track the color parameters until an appropriate extraction result is obtained while viewing the confirmation area 135 and the binary image.
Japanese Patent Laid-Open No. 2-78937 JP 9-145633 A

  The board inspection apparatus has an advantage that a plurality of inspection items can be inspected at high speed and accurately at once for the mounting quality of the printed board. However, in actual operation of the board inspection device, the over logic that teaches each inspection logic according to the individual inspection object, does not miss the defective product, and determines that the non-defective product is a defective product is an allowable value ( Judgment accuracy must be sufficiently increased until the value can be kept below the value assumed in advance.

  Complementing the oversight and overdetection of defective products, the outflow of defective products is never allowed in any inspection. If it is difficult to judge defective products and non-defective products, it is better to judge non-defective products as defective products. However, excessive detection increases the loss cost of discarding non-defective products as defective products or requires re-inspection of defective products. The merit of automating the inspection is lost.

  However, in the color highlight type substrate inspection apparatus, although advanced substrate inspection that can withstand practical use is possible, teaching to suppress oversight and overdetection of defective products to a target value is difficult. Even if the color parameter setting support tool described above is used, it is inevitable that the setting of a color parameter will depend on the experience and intuition of the operator. In addition, no matter how good the operator is, the adjustment must be repeated on a trial and error basis, which is inefficient and requires a lot of labor and adjustment time.

  In a manufacturing environment where the product life cycle is shortening and the manufacturing environment is changing rapidly, it is strongly desired to reduce teaching work and to further automate teaching.

  The present invention has been made in view of the above circumstances, and an object of the present invention is to provide a technique capable of automatically generating parameters used in a substrate inspection apparatus.

  In the present invention, an information processing device (parameter setting device) is obtained by first imaging a plurality of target images obtained by imaging a component to be detected by inspection and a component to be excluded by inspection. Acquire multiple excluded images. For example, in the case of a non-defective product inspection for detecting a non-defective product, an image of a component with a good mounting state is a target image, and an image of a component with a poor mounting state is an excluded image. Conversely, in the case of a defect inspection for detecting a defect, an image of a part having the defect is a target image, and other images are excluded images.

  Subsequently, the information processing apparatus maps the color of each pixel of the target image as a target point and the color of each pixel of the excluded image as an excluded point to the color space. Then, the information processing apparatus obtains a color range that divides the color space so that a difference between the number of target points included in the color space and the number of exclusion points is maximized, and the obtained color The range is set as a color condition (color parameter) used in substrate inspection. Thereby, a color condition which is one of inspection parameters is automatically generated.

  Here, the color space may be a multi-dimensional color space consisting of at least three axes of brightness, hue, and saturation. For example, the color space tends to be included in the target image and hardly included in the excluded image. It is also preferable to use a two-dimensional color space composed of a saturation axis and a brightness axis for hue. Employing the two-dimensional color space simplifies the color range search process. In the two-dimensional color space, if the color condition is composed of a lower limit and an upper limit of saturation and a lower limit and an upper limit of lightness, the color range becomes a rectangular area, and the color range search process is further simplified.

  Subsequently, the information processing apparatus extracts a pixel region that satisfies the color condition from each of the target image and the excluded image, creates a feature amount histogram for a feature amount of the pixel region, and the feature amount histogram Calculating a threshold for separating the feature amount of the target image and the feature amount of the excluded image based on the frequency distribution of the target image, and setting the calculated threshold value as a determination condition used in substrate inspection Good. Thereby, a determination condition (threshold value), which is one of inspection parameters, is automatically generated.

  Here, various features such as the area, area ratio, length, maximum width, center of gravity, and shape of the pixel region are assumed as the feature amount. What is necessary is just to employ | adopt one or two or more preferable feature-values according to the object which should be detected by test | inspection.

  Each of the above processes is executed by a program of the information processing apparatus. The automatically generated parameters (color conditions and determination conditions) are stored in the storage unit of the substrate inspection apparatus and used for substrate inspection processing.

  The board inspection apparatus irradiates a mounting component on the board with light of a plurality of colors at different incident angles, and extracts an area that satisfies the color condition from an image obtained by imaging the reflected light. The mounting state of the component is inspected based on whether or not the feature quantity of the region satisfies the determination condition.

  By the way, the color of the target image may vary depending on the position of the component and the lighting condition. In this case, target points (outliers) that are out of the color range increase, and pixel areas are not appropriately extracted, which may lead to a decrease in determination accuracy.

  Therefore, when an outlier exists in the frequency distribution of the target image in the feature amount histogram, a second color range is searched using the target image corresponding to the outlier and the plurality of excluded images, and the color A set sum of the range and the second color range may be set as a color condition used in the substrate inspection. According to this new color condition, in addition to the target point extracted under the original color condition, an outlier target point can also be extracted, so that the target image and the excluded image can be appropriately separated.

  Further, when an outlier exists in the frequency distribution of the target image in the feature amount histogram, the target image corresponding to the outlier and a plurality of excluded images are searched for a second color range, and the A third color range is searched using a target image that is not an outlier and the plurality of excluded images, and a set sum of the second color range and the third color range is set as a color condition used in substrate inspection. It is also preferable. Also by this method, target points of outliers can be extracted, and target images and excluded images can be appropriately separated.

  Further, the information processing apparatus divides the target points mapped in the color space into a plurality of groups based on the color distribution, searches for a color range for each group, and performs a substrate inspection on a set sum of the color ranges of all the groups. It is also preferable to set the color condition used in the above. Also by this method, the outlier target points can be extracted, and the target image and the excluded image can be appropriately separated.

  Note that the present invention can be understood as a parameter setting method for a substrate inspection apparatus including at least a part of the above processing, or a program for realizing the method. The present invention can also be understood as a parameter setting device of a substrate inspection apparatus having at least a part of the means for executing the above processing, or a substrate inspection apparatus equipped with such a device. Each of the above means and processes can be combined with each other as much as possible to constitute the present invention.

  According to the present invention, parameters used in a substrate inspection apparatus can be automatically generated, teaching work can be reduced, and teaching can be automated.

  Exemplary embodiments of the present invention will be described in detail below with reference to the drawings.

<First Embodiment>
(Configuration of board inspection system)
FIG. 1 shows a hardware configuration of a board inspection system according to an embodiment of the present invention.

  The substrate inspection system includes a substrate inspection device 1 that executes substrate inspection processing and a parameter setting device 2 that automatically generates parameters used in the substrate inspection processing of the substrate inspection device 1. The board inspection apparatus 1 and the parameter setting apparatus 2 can exchange electronic data such as images and parameters via a wired or wireless network, or a recording medium such as an MO or a DVD. In the present embodiment, the board inspection apparatus 1 and the parameter setting apparatus 2 are configured separately, but it is also possible to integrate the function of the parameter setting apparatus into the board inspection apparatus main body.

(Configuration of board inspection equipment)
The board inspection apparatus 1 is an apparatus for automatically inspecting the mounting quality (soldering state, etc.) of the mounting component 21 on the board 20 by a color highlight method. The substrate inspection apparatus 1 generally includes an X stage 22, a Y stage 23, a light projecting unit 24, an imaging unit 25, and a control processing unit 26.

  Each of the X stage 22 and the Y stage 23 includes a motor (not shown) that operates based on a control signal from the control processing unit 26. By driving these motors, the X stage 22 moves the light projecting unit 24 and the imaging unit 25 in the X axis direction, and the Y stage 23 moves the conveyor 27 that supports the substrate 20 in the Y axis direction.

The light projecting unit 24 includes three annular light sources 28, 29, and 30 that have different diameters and that simultaneously emit red light, green light, and blue light based on a control signal from the control processing unit 26. . The light sources 28, 29, and 30 are arranged in directions corresponding to different elevation angles when centered on the position directly above the observation position and viewed from the observation position. With this arrangement, the light projecting unit 24 irradiates the mounting component 21 on the substrate 20 with light of a plurality of colors (in this embodiment, three colors of R, G, and B) at different incident angles.

  The imaging unit 25 is a color camera, and is positioned downward at a position directly above the observation position. As a result, the reflected light on the substrate surface is imaged by the imaging unit 25, converted into color signals R, G, and B of the three primary colors and supplied to the control processing unit 26.

  The control processing unit 26 includes an A / D conversion unit 33, an image processing unit 34, an inspection logic storage unit 35, a determination unit 36, an imaging controller 31, an XY stage controller 37, a memory 38, a control unit (CPU) 39, and a storage unit 32. , An input unit 40, a display unit 41, a printer 42, a communication I / F 43, and the like.

  The A / D conversion unit 33 is a circuit that receives the color signals R, G, and B from the imaging unit 25 and converts them into digital signals. The digital grayscale image data for each hue is transferred to the image data storage area in the memory 38.

  The imaging controller 31 is a circuit including an interface for connecting the control unit 39 to the light projecting unit 24 and the image capturing unit 25, and the light amount of each light source 28, 29, 30 of the light projecting unit 24 based on the output of the control unit 39. And control such as maintaining the mutual balance of the hue light outputs of the imaging unit 25 is performed.

  The XY stage controller 37 is a circuit including an interface for connecting the control unit 39 to the X stage 22 and the Y stage 23, and controls driving of the X stage 22 and the Y stage 23 based on the output of the control unit 39.

  The inspection logic storage unit 35 is a storage unit that stores inspection logic used for substrate inspection processing. The board inspection apparatus 1 can perform a plurality of types of inspection processes such as a fillet inspection for inspecting a solder shape and a lack inspection for inspecting a missing part. The inspection logic is prepared for each type of inspection, and includes color parameters (color conditions) for extracting a predetermined color pattern (pixel area) from the image, types of feature values extracted from the color pattern, Consists of pass / fail judgment conditions related to the feature amount.

  The image processing unit 34 performs a process of extracting a region satisfying the color parameter from an image obtained by imaging the component 21 on the substrate 20 and a process of calculating a predetermined feature amount from the extracted region. It is. The determination unit 36 is a circuit that receives the feature amount calculated by the image processing unit 34 and executes processing for determining whether the component mounting state is good or not based on whether or not the feature amount satisfies a predetermined determination condition.

  The input unit 40 includes a keyboard, a mouse, and the like necessary for inputting operation information and data related to the substrate 20. The input data is supplied to the control unit 39. The communication I / F 43 is for transmitting and receiving data to and from the parameter setting device 2 and other external devices.

  The control unit (CPU) 39 is a circuit that executes various arithmetic processes and control processes. The storage unit 32 is a storage device composed of a hard disk and a memory, and stores the CAD information of the substrate, the determination result of the substrate inspection process, and the like in addition to the program executed by the control unit 39.

FIG. 2 shows a functional configuration of the substrate inspection apparatus 1. The substrate inspection apparatus 1 includes an instruction information receiving function 10,
A board carry-in function 11, a CAD information read function 12, a stage operation function 13, an imaging function 14, an inspection logic read function 15, an inspection function 16, a determination result writing function 17, and a board carry-out function 18 are provided. These functions are realized by the control unit 39 controlling the hardware according to a program stored in the storage unit 32. In addition, a CAD information storage unit 32 a that stores CAD information and a determination result storage unit 32 b that stores determination results are provided inside the storage unit 32.

(Board inspection processing)
Next, substrate inspection processing in the substrate inspection apparatus 1 will be described. Here, fillet inspection will be described as an example of substrate inspection processing. Fillet inspection is processing for determining whether or not the shape of a solder fillet is good.

  As shown in the upper part of FIG. 3, in the good solder fillet, a wide inclined surface like a mountain skirt is formed from the component 21 to the land on the substrate 20. On the other hand, when the solder shortage occurs, the area of the inclined surface is reduced. Conversely, when the solder is excessive, the solder fillet is raised on the land.

  When these solder fillets are imaged by the substrate inspection apparatus 1, images as shown in the middle of FIG. 3 are obtained. Since the red, green, and blue irradiation lights are incident on the solder fillet at different angles, the hue of the reflected light that is incident on the imaging unit 25 changes according to the inclination of the solder fillet. That is, the reflected light of the blue light having the shallowest incident angle is dominant in the steep portion, whereas the reflected light of the red light is dominant in the portion having almost no inclination. Therefore, the region of the blue hue is large in the non-defective solder fillet, and the region of the hue other than blue is large in the defective solder fillet.

  In the fillet inspection according to the present embodiment, the quality of the solder fillet is determined based on the size (area) of the blue region using such a tendency of the color pattern. Hereinafter, the flow of fillet inspection processing will be described in detail with reference to the flowchart of FIG.

  The instruction information receiving function 10 is in a waiting state until instruction information for instructing execution of substrate inspection is input (step S100; NO, step S101). When instruction information is input from an external device by operating the input unit 40 or via the communication I / F 43, the instruction information receiving function 10 receives the instruction information, the board carry-in function 11, the CAD information reading function 12, and the inspection logic. The data is sent to the reading function 15 (step S100; YES). The instruction information includes information (model number, etc.) of the board to be inspected.

  The inspection logic reading function 15 reads the inspection logic corresponding to the model number of the board from the inspection logic storage unit 35 (step S102). Here, the inspection logic for the fillet inspection is read. The inspection logic includes a color parameter (color condition) and a threshold value (determination condition).

  The board carry-in function 11 carries the board 20 to be inspected from the printed board carry-in section onto the conveyor 27 based on the instruction information (step S103), and the CAD information reading function 12 loads the CAD corresponding to the board model number. Information is read from the CAD information storage unit 32a (step S103).

  Next, the stage operation function 13 obtains information such as the dimensions, shape, and component arrangement of the substrate 20 from the read CAD information, and the plurality of components 21 mounted on the substrate 20 are sequentially observed positions (imaging positions). The X stage 22 and the Y stage 23 are operated via the XY stage controller 37 (step S105).

  On the other hand, the imaging function 14 causes the three light sources 28, 29, and 30 of the light projecting unit 24 to emit light via the imaging controller 31, and simultaneously irradiates red, green, and blue light onto the substrate 20. In addition, the imaging function 14 controls the imaging unit 25 via the imaging controller 31, and images the component 21 on the substrate 20 in synchronization with the operation of the stages 22 and 23 (step S106). The captured image is taken into the memory 38.

  Next, the inspection function 16 extracts a solder region from the captured image by the image processing unit 34 (step S107). The extraction of the solder region can be automatically performed by template matching, for example.

  Subsequently, the inspection function 16 binarizes the extracted solder region using the color parameter (step S108). The color parameters used here are composed of four values: a lower limit and an upper limit of blue saturation, and a lower limit and an upper limit of lightness. In the binarization process, pixels included in the color range defined by the color parameter are converted to white pixels, and other pixels are converted to black pixels.

  The lower part of FIG. 3 shows the solder area after binarization. By binarizing with the color parameter, it can be seen that only the blue color region in the solder region is extracted as a white pixel, and the difference (feature) between the non-defective image and the defective product image is clarified.

  Subsequently, the inspection function 16 uses the image processing unit 34 to extract the feature amount of the white pixel region. Here, the area (number of pixels) of the white pixel region is calculated as the feature amount. Then, the inspection function 16 passes the area value of the white pixel region to the determination unit 36, and the determination unit 36 compares the area value of the white pixel region with a threshold value (step S109). When the area value exceeds the threshold value (step S109; YES), it is determined that the solder mounting quality of the component 21 is good (step S110), and when the area value is less than the threshold value (step S109). NO), it is determined that the solder mounting quality of the component 21 is poor (step S111).

  The determination result writing function 17 writes the determination result together with the location ID (information for specifying a part) in the determination result storage unit 32b (step S112).

  When all the components on the board 20 have been inspected, the board carry-out function 18 carries out the board 20 by the printed board carrying section, and the board inspection process is finished (step S113).

  According to the board inspection processing described above, since the three-dimensional shape of the solder fillet can be accurately grasped by the color pattern appearing in the two-dimensional image, it is possible to accurately determine the quality of the solder mounting quality.

  By the way, in order to achieve high determination accuracy so that defective products are not overlooked and overdetection is less than an allowable value, the color parameters (color conditions) and threshold values (determination conditions) of the inspection logic are set in advance. It is necessary to set the optimum value according to the inspection object. In the present embodiment, the generation (teaching) of these parameters is automatically performed by the parameter setting device 2. This will be described in detail below.

(Configuration of parameter setting device)
As shown in FIG. 1, the parameter setting device 2 is a general-purpose device including a CPU, a memory, a hard disk, an I / O control unit, a communication I / F, a display unit, an information input unit (keyboard and mouse), etc. as basic hardware. It is composed of a computer (information processing device).

FIG. 5 shows a functional configuration of the parameter setting device 2. The parameter setting device 2 includes an instruction information reception function 50, a teacher image information reading function 51, an image acquisition function 52, a distribution function 53, a mapping function 54, a color range search function 55, a binarization function 56, and a feature amount histogram generation function. 57, a threshold value determination function 58, an inspection logic generation function 59, and an inspection logic writing function 60. These functions are realized by a program stored in a memory or a hard disk being read and executed by a CPU.

  In addition, a teacher image information DB 61 for storing teacher image information used for teaching is provided in the hard disk. The teacher image information includes an image of the mounted component imaged by the board inspection apparatus 1 and teacher information (teaching data) indicating whether the image should be detected (non-defective product) or excluded (defective product). In order to increase the reliability of teaching, it is preferable to prepare dozens to thousands of teacher image information for each of the non-defective product and the defective product.

(Parameter setting process)
The flow of parameter setting processing will be described with reference to the flowchart of FIG. In the present embodiment, an example of generating inspection parameters used in the above-described fillet inspection will be described.

  The instruction information receiving function 50 is in a waiting state until instruction information for instructing automatic generation of inspection logic is input (step S200; NO, step S201). When the instruction information is input from the information input unit, the instruction information receiving function 50 transmits the instruction information to the teacher image information reading function 51 (step S200; YES). This instruction information includes information for specifying teacher image information to be subjected to inspection logic generation, the type of inspection logic, and the like.

  The teacher image information reading function 51 reads teacher image information corresponding to the examination logic to be created from the teacher image information DB 61 in accordance with the instruction information (step S202). The teacher image information includes a non-defective image (an image showing a solder fillet having a good shape) and a defective product image (an image showing a solder fillet having a bad shape). Teacher information is given to these images.

  FIG. 7 shows an example of a good product image and a defective product image. In the good product image, good solder fillets are formed in the land regions 63 and 63 at both ends of the component 62. Teacher information “good” is assigned to each of the land areas 63 and 63. On the other hand, in the defective product image, the component 64 is mounted with an inclination, and a solder failure occurs in the land region 65 on one side. Therefore, teacher information “defective” is given to the land area 65. Note that teacher information “ignore” is assigned to the land area 66 on the opposite side.

  When the teacher image information is read, the image acquisition function 52 extracts the solder area to which the teacher information is added from the teacher image information (step S203). As shown in FIG. 8, the image acquisition function 52 has a template composed of a land window 70 and a component main body window 71, and enlarges / reduces the template, or relative to the land window 70 and the component main body window 71. The windows 70 and 71 are aligned with the land regions 63 and 65 and the parts 62 and 64 in the image while shifting the position. For example, a method such as template matching may be used for matching the windows. Thereby, the land areas 63, 65, and 66 are specified for the non-defective product image and the defective product image, respectively. And the area | region except the overlap part with the component main body window 71 from the land window 70 is extracted as a solder area | region (refer the shaded part of FIG. 8). Note that the land area itself (the entire land window 70) may be handled as a solder area.

Next, the distribution function 53 distributes the extracted solder area into the target image and the excluded image based on the teacher information (step S204). In this example, since the non-defective product detection is intended, the solder area to which the teacher information “good” is given is set as the target image, and the solder area to which the teacher information “bad” is given is set as the excluded image. The solder area to which the teacher information “ignore” is given is ignored.

  The target image extracted here represents a good solder fillet, and the excluded image represents a bad solder fillet. Therefore, creating optimal color parameters for fillet inspection involves finding an optimal solution in a color range that includes as much of the pixel color of the target image as possible and can almost eliminate the color of the pixel of the excluded image. Is equivalent to

  Therefore, first, the mapping function 54 maps the colors of all pixels of the target image and the excluded image to the color histogram (step S205). At this time, the pixel of the target image is “target point”, and the pixel of the excluded image is “excluded point”, and the mapping is performed in a mutually distinguishable format. The color histogram is obtained by recording the frequency (number) of pixels at each point in the color space. The color histogram can grasp the color distribution of the pixels constituting the solder area. Note that the pixel referred to here is the minimum resolution of an image. When the mapping process is executed collectively for a plurality of pixels, color mixing occurs, and therefore the process for each pixel is preferable.

  In general, the color space is composed of a multidimensional space of at least hue, saturation, and brightness. Therefore, in order to accurately grasp the color distribution of the pixel, it is preferable to use a multidimensional color histogram in which the color of the pixel is mapped in the multidimensional color space.

  However, in the color highlight method, since red and blue are used for the light source, a red or blue color tends to appear strongly in the solder area (this causes reflection close to specular reflection on the solder surface). Because.) Further, as described above, it is also known that most pixels have a blue color in a good solder region, and most pixels have a red color in a defective solder region.

  Therefore, if the purpose is to determine the color parameters for separating the non-defective product color (target point) and the defective product color (exclusion point), one color (for example, blue) or two colors (for example, blue and red) are considered. This is enough. Therefore, in the present embodiment, blue is selected as a hue that is included in the target image and tends to be hardly included in the excluded image, and the pixel is placed in a two-dimensional color space including a blue saturation axis and a brightness axis. A two-dimensional color histogram in which colors are mapped is used. This greatly simplifies the algorithm for finding the optimal color parameter solution.

  FIG. 9 shows an example of a two-dimensional color histogram. The horizontal axis represents blue saturation. The larger the positive value, the stronger the blue component, and the larger the negative value, the stronger the yellow component, which is the complementary color of blue. The vertical axis represents the brightness, and the brightness increases as the value increases. White circles (◯) in the histogram represent target points, and black triangles (▲) represent exclusion points. It can be seen that there is a difference in color distribution between the target point and the excluded point.

  Next, the color range search function 55 searches for a color range that optimally divides the color distribution of the target point and the color distribution of the exclusion point based on the two-dimensional color histogram (step S206). In this embodiment, for simplification of the algorithm, as shown in FIG. 10A, the lower limit (BInf) and upper limit (BSup) of saturation, and the lower limit (LInf) and upper limit (LSup) of lightness are included. Consider a rectangular color range. The optimum solution to be obtained here is a color range that includes as many target points (◯) as possible and hardly includes exclusion points (（).

Specifically, the color range search function 55 calculates the frequency total value E for each color range while changing the values of BInf, BSup, LInf, and LSup (see Equation 1), and the frequency total value E is maximized. Find the color range. The frequency total value E is an index representing the difference between the number of target points (frequency) included in the color range and the number of excluded points (frequency). FIG. 10B shows a color range in which the frequency total value E is maximum.

  Then, the color range search function 55 sets a color range in which the frequency total value E is maximized as an inspection color parameter (color condition). As described above, according to the present embodiment, it is possible to automatically generate color parameters for appropriately separating the target image (target point) and the excluded image (excluded point).

  Next, processing for automatically generating an inspection threshold value (determination condition) is executed using the color parameter.

  First, the binarization function 56 binarizes all the solder areas of the non-defective product image and the defective product image using the color parameter (step S207). In this binarization processing, pixels included in the color range defined by the color parameter are converted to white pixels, and other pixels are converted to black pixels.

  As shown in FIG. 11, the white pixel region is very large in the non-defective image, and the white pixel region is extremely small in the defective image. Therefore, when such a binarized image is used, it becomes easy to quantitatively calculate a feature amount for identifying a non-defective product or a defective product. Examples of the feature amount include the area, area ratio, center of gravity, length, maximum width, and shape of the white pixel region. Here, the area is selected as the feature amount.

  The feature amount histogram generation function 57 grasps the difference between the distribution trend of the feature amount of the non-defective image and the distribution trend of the feature amount of the non-defective image, and thus the area value of the white pixel region for each of the non-defective image and the defective product image. An area histogram is created for step S208. FIG. 12 shows an example of an area histogram of a non-defective product image and a defective product image (hereinafter simply referred to as “good product histogram” or “defective product histogram”). It can be seen that a clear difference appears between the feature quantity distribution of the non-defective image and the feature quantity distribution of the defective product image.

  Next, the threshold value determination function 58 calculates a threshold value for optimally separating the feature quantity of the non-defective product image and the feature quantity of the defective product image based on the frequency distribution of the non-defective product histogram and the defective product histogram ( Step S209). Various methods for optimally separating the two peaks appearing in the feature amount histogram have been proposed, and any method may be adopted here. For example, Otsu's discriminant analysis method may be used, or a point separated by 3σ from the edge of a non-defective image peak may be determined as a threshold based on experience. In this way, a threshold value for discriminating between non-defective products and defective products is generated.

Then, the inspection logic generation function 59 generates inspection logic from the color parameter, the type of feature amount (area in this example), and the threshold value (step S210), and the inspection logic writing function 60 generates the inspection logic on the substrate. The data is written in the inspection logic storage unit 35 of the inspection device 1 (step S211).

  According to the present embodiment described above, the inspection logic (parameter) used in the substrate inspection process is automatically generated, so that the time and load required for teaching can be greatly reduced.

  In addition, since the optimum color parameter and threshold value are calculated by the above-described algorithm, it is possible to perform pass / fail judgment by the color highlight method with high accuracy. Note that the reliability of the color parameter and the threshold value is improved as the number of teacher image information to be given first increases.

Second Embodiment
Depending on the position of the parts and lighting conditions, the color of the image may vary. For example, an image with a small blue component may be obtained even if it is a non-defective product due to the influence of shadows of other parts. At this time, an outlier appears in the distribution of the target points on the color histogram (see FIG. 13A).

  When a color range search is executed for a color histogram in which such outliers exist using the same method as in the first embodiment, as shown in FIG. 13B, target points that are outliers are included. A rectangular area with no rectangle may be chosen as the optimal solution. This is because if the rectangular range is set so as to include all the target points as shown in FIG. 13C, many exclusion points are included, and the frequency total value E decreases.

  Then, when the binarization process is executed with color parameters that do not include outliers, as shown in FIG. 14, a non-defective image that includes many outliers hardly forms a white pixel area, and thus the frequency distribution of the nondefective histogram varies. Arise. In such a case, there is an overlap between the good product histogram and the defective product histogram, and it becomes impossible to properly separate the good product and the defective product.

  Therefore, in the second embodiment of the present invention, parameter setting processing considering outliers is performed with the following configuration.

  FIG. 15 shows a functional configuration of the parameter setting device 2. The parameter setting device 2 of this embodiment includes an outlier determination function 80, an outlier extraction function 81, and a color range combination function 82 in addition to the functions of the first embodiment (see FIG. 5). These functions are also realized by a program stored in a memory or a hard disk being read and executed by a CPU.

  The flow of the parameter setting process in the second embodiment will be described along the flowchart of FIG. Note that the same steps as those in the first embodiment are denoted by the same step numbers.

  In the parameter setting device 2, first, color parameters are generated according to the same processing as in the first embodiment (steps S200 to S206), and a non-defective product histogram and a defective product histogram are generated using the color parameters (steps S207 to S208). ).

  Here, the outlier determination function 80 determines the presence or absence of an outlier from the good product histogram and the defective product histogram (step S300). Specifically, when there is an overlap between the good product histogram and the defective product histogram, it is determined that “there is an outlier”, and when there is no overlap, it is determined that there is no “outlier”.

If there is an outlier (step S301; YES), the outlier extraction function 81 separates the outlier portion from the non-defective histogram (step S302), and a target image corresponding to the outlier (non-defective solder region). ) Only (step S303).

  Then, the mapping function 54 maps the color of all pixels of the target image extracted in step S303 and the color of all pixels of the excluded image to the color histogram in the same manner as in step S205 (step S304). The color histogram obtained here includes only the color distribution of excluded points and the color distribution of target points of outliers, as shown in FIG. Therefore, when the color range search function 55 searches for a color range based on this color histogram, a second color range is obtained that suitably separates outlier target points as shown in FIG. S305).

  Next, the color range combination function 82 adds the newly searched second color range to the color parameter (step S306). At this time, the new color parameter is set to the set sum of the already searched first color range and the second color range obtained this time (see FIG. 18). That is, in the binarization process, if a pixel is included in any one of a plurality of color ranges constituting the color parameter, the pixel is converted into a white pixel.

  The pixels of the target image (non-defective image) that are not outliers are out of the second color range, but are included in the first color range. Therefore, the white pixel area is converted into a binary image. Further, pixels of the outlier target image (non-defective product image) are out of the first color range, but are included in the second color range. Therefore, an outlier target image is also converted into a binarized image in which the white pixel region occupies most. On the other hand, in the case of an excluded image (defective product image), most of the pixels out of the first color range and the second color range occupy the white pixel area of the binarized image.

  After the color parameter is updated, binarization of the solder area and creation of an area histogram are performed using the color parameter (steps S207 and S208), and the presence or absence of outliers is verified (step S300). If the outlier still exists, the second color range is searched and the color parameter is updated again using the outlier (steps S302 to S306). In this case, the color parameter is a set sum of three color ranges.

  The above process is repeated until there is no outlier (step S301). When the color parameter is determined (step S301; NO), the threshold value is determined in the same manner as in the first embodiment, inspection logic is generated and recorded (steps S209 to S211), and the parameter setting process ends.

  According to the present embodiment described above, even when there is a variation (outlier) in the color of a non-defective image, an optimal color parameter that can appropriately separate a non-defective product from a defective product is automatically generated. Therefore, the determination accuracy can be further improved.

<Third Embodiment>
Also in the third embodiment of the present invention, parameter setting processing considering outliers is performed. However, in the second embodiment, the color range is re-searched only for the outlier solder area, whereas in this embodiment, the color range is similarly searched for the non-outlier solder area. Different.

  The flow of parameter setting processing in the third embodiment will be described along the flowchart of FIG. In addition, the same step number is attached | subjected about the process same as 1st or 2nd embodiment.

In the parameter setting device 2, first, color parameters are generated according to the same processing as in the second embodiment (steps S200 to S206), and a non-defective product histogram and a defective product histogram are generated using the color parameters (steps S207 to S208). The presence or absence of outliers is determined (step S300).

  When an outlier exists (step S301; YES), the outlier extraction function 81 divides the good product histogram into an outlier portion and a non-outlier portion (step S310), and a target image corresponding to the outlier. And target images that are not outliers are extracted separately (step S311).

  The mapping function 54 first maps the colors of all pixels of the outlier target image and the colors of all pixels of the excluded image to the first color histogram. Subsequently, the mapping function 54 maps the colors of all the pixels of the target image that are not outliers and the colors of all the pixels of the excluded image to the second color histogram (step S312). As shown in FIG. 20, the first color histogram includes only the color distribution of the excluded points and the color distribution of the outlier target points, and the second color histogram includes the color distribution of the excluded points and the target points that are not outliers. Only the color distribution is included.

  The color range search function 55 searches for a color range separately in each color histogram (step S313). Then, a second color range is obtained from the first color histogram so as to suitably cut out the target points of outliers, and the third color range is suitably cut out from the second color histogram. Can be obtained.

  The color range combination function 82 sets a set sum of the second color range and the third color range as a new color parameter (step S314). The subsequent processing is the same as in the second embodiment.

  According to the present embodiment described above, even when there is a variation (outlier) in the color of a non-defective image, an optimal color parameter that can appropriately separate a non-defective product from a defective product is automatically generated. Therefore, the determination accuracy can be further improved. In addition, since the color range is searched again for target points that are not outliers, the reliability of the color parameters is further increased.

<Fourth embodiment>
Also in the fourth embodiment of the present invention, parameter setting processing is performed in consideration of outliers. However, in the second and third embodiments, the presence / absence of outliers is determined based on the area histogram, whereas in this embodiment, the presence / absence of outliers is determined based on the color histogram.

  FIG. 21 shows a functional configuration of the parameter setting device 2. The parameter setting device 2 of the present embodiment has a grouping function 83 as a new function. Other functions are the same as those of the above embodiment.

  The flow of parameter setting processing in the fourth embodiment will be described along the flowchart of FIG. Note that the same steps as those in the first embodiment are denoted by the same step numbers.

  In the parameter setting device 2, first, a color histogram is created according to the same processing as in the first embodiment (steps S200 to S205).

  Next, the grouping function 83 determines whether or not an outlier exists at the target point mapped to the color histogram (step S320). The presence or absence of outliers can be determined by using, for example, a general clustering method.

When there is no outlier (that is, in the case of a single class) (step S321; NO), color parameters and threshold values are generated in the same manner as in the first embodiment (steps S206 to S211). .

  On the other hand, when there is an outlier (that is, when it is classified into a plurality of classes) (step S321; YES), the grouping function 83 groups the target points for each class (step S322). Then, the color range search function 55 searches the first color range from the target points belonging to the first group and all the excluded points, and the second from the target points belonging to the second group and all the excluded points. In the same manner, the color range is searched for each group (step S323).

  Then, the color range combining function 82 sets the set sum of the color ranges of all groups as the color parameter (step S324). The subsequent processing is the same as in the first embodiment.

  According to the present embodiment described above, even when there is a variation (outlier) in the color of a non-defective image, an optimal color parameter that can appropriately separate a non-defective product from a defective product is automatically generated. Therefore, the determination accuracy can be further improved. In addition, since the target points are grouped based on the color distribution and the color ranges are individually searched, the reliability of the color parameters is further increased.

  In the present embodiment, grouping is performed in units of target points, but grouping may be performed in units of parts (target images). In this case, for example, the color centroid is obtained for each target image, mapped to a color histogram, and clustering processing is performed based on the distribution of the color centroid, thereby determining the presence or absence of a component image that becomes an outlier.

<Modification>
The first to fourth embodiments described above are merely examples of the present invention. The scope of the present invention is not limited to the above embodiment, and various modifications can be made within the scope of the technical idea.

  For example, in the above embodiment, a two-dimensional color histogram (color space) is used, but a multi-dimensional (hue, saturation, brightness) color histogram may be used. Also for the two-dimensional color histogram, the saturation axis of other hues such as red, green, and yellow may be used instead of the blue saturation axis, or the hue axis may be used instead of the saturation axis. The selection of the axis of the color histogram may be determined according to the tendency of the color pattern of the component image captured by the board inspection apparatus.

  Further, the color range is not limited to a rectangle, and a circle, a polygon, a free curve figure, or the like may be used. Furthermore, when the color histogram is multidimensional, the color range may be multidimensional.

  In the above embodiment, the fillet inspection has been described as an example. However, the present invention performs area extraction based on color parameters (color conditions), and determines some feature amount of the extracted area based on the determination conditions. If so, the present invention can be applied to other substrate inspection processes.

  In the above embodiment, the area is used as the feature amount. However, as the feature amount used for the pass / fail judgment, an area ratio, a length, a maximum width, a center of gravity, and the like can be preferably used. The area ratio is the occupation ratio of the binarized area in the land window. For example, if a component is soldered with a deviation from the land area, the area of the solder area increases and the area ratio changes. If this is regarded as a feature amount, it is effective for inspection of component displacement. The length is the length in the vertical direction or the horizontal direction of the white pixel region, and the maximum length is the maximum value among the lengths of the white pixel region. The center of gravity is a relative position of the center of gravity of the white pixel region with respect to the land window.

Any feature amount may be used as long as it can accurately determine pass / fail, and it is also preferable to combine a plurality of types of feature amounts to improve accuracy. It is also possible to extract a plurality of types of feature amounts in the parameter setting process, and to adopt a feature amount that best separates non-defective products and defective products among them. In the above embodiment, an area histogram (area value histogram) is used to determine the determination condition (threshold value). However, if the type of feature amount is different, a feature amount histogram (area ratio histogram, length histogram, Maximum width histogram, centroid histogram, etc.). For example, if an area ratio histogram is used instead of the area value histogram, the pass / fail judgment is executed based on the pixel occupation rate binarized by the color parameter in the land window. Even when the size becomes smaller or larger, determination processing that is not affected by the size of the land window is possible.

The figure which shows the hardware constitutions of the board | substrate inspection system which concerns on embodiment of this invention. The figure which shows the function structure of a board | substrate inspection apparatus. The figure which shows the relationship between the shape of a solder fillet, an imaging pattern, and a binarized image. The flowchart which shows the flow of a board | substrate inspection process. The figure which shows the function structure of the parameter setting apparatus which concerns on 1st Embodiment. The flowchart which shows the flow of the parameter setting process which concerns on 1st Embodiment. The figure which shows an example of a quality product image and a defective product image. The figure which shows the extraction process of a solder area | region. The figure which shows an example of a two-dimensional color histogram. The figure which shows the search process of a color range. The figure which shows an example of the binarization result of a non-defective product image and a defective product image. The figure which shows the area histogram and threshold value determination process of a good article and inferior goods. The figure which shows an example of the color histogram containing an outlier. The figure which shows an area histogram in case an outlier exists. The figure which shows the function structure of the parameter setting apparatus which concerns on 2nd Embodiment. The flowchart which shows the flow of the parameter setting process which concerns on 2nd Embodiment. The figure which shows the search process of the 2nd color range for carving out the target point of an outlier. The figure which shows the color parameter production | generation process in 2nd Embodiment. The flowchart which shows the flow of the parameter setting process which concerns on 3rd Embodiment. The figure which shows the color parameter production | generation process in 3rd Embodiment. The figure which shows the function structure of the parameter setting apparatus which concerns on 4th Embodiment. The flowchart which shows the flow of the parameter setting process which concerns on 4th Embodiment. The figure which shows the structure of the board | substrate inspection apparatus of a color highlight system. The figure which shows an example of the color pattern which appears in a captured image. The figure which shows the setting support tool of a color parameter.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Board | substrate inspection apparatus 2 Parameter setting apparatus 10 Instruction information reception function 11 Board | substrate carrying-in function 12 CAD information reading function 13 Stage operation function 14 Imaging function 15 Inspection logic reading function 16 Inspection function 17 Judgment result writing function 18 Board | substrate unloading function 20 Board | substrate 21 Mounted parts 22 X stage 23 Y stage 24 Projection unit 25 Imaging unit 26 Control processing unit 27 Conveyor 28 Red light source 29 Green light source 30 Blue light source 31 Imaging controller 32 Storage unit 32a CAD information storage unit 32b Judgment result storage unit 33 A / D Conversion unit 34 Image processing unit 35 Inspection logic storage unit 36 Determination unit 37 XY stage controller 38 Memory 39 Control unit 40 Input unit 41 Display unit 42 Printer 50 Instruction information reception function 51 Teacher image information reading function 52 Image acquisition function 53 Distribution function 4 Mapping Function 55 Color Range Search Function 56 Binarization Function 57 Feature Quantity Histogram Generation Function 58 Threshold Determination Function 59 Inspection Logic Generation Function 60 Inspection Logic Write Function 62, 64 Parts 63, 65, 66 Land Area 70 Land Window 71 Component body window 80 Outlier determination function 81 Outlier extraction function 82 Color range combination function 83 Grouping function

Claims (12)

A region satisfying a predetermined color condition is extracted from an image obtained by irradiating a component mounted on the board with light of multiple colors at different angles of incidence and imaging the reflected light, and the features of the extracted region Is a method of automatically generating parameters used in a board inspection apparatus that inspects the mounting state of the component depending on whether or not a predetermined determination condition is satisfied,
Information processing device
Obtaining a plurality of target images obtained by imaging the parts to be detected by the inspection and a plurality of exclusion images obtained by imaging the parts to be excluded by the inspection;
Mapping the color of each pixel of the target image as a target point and the color of each pixel of the excluded image as an exclusion point, respectively, and mapping them to a color space,
A color range that divides the color space, and obtaining a color range that maximizes the difference between the number of target points included therein and the number of excluded points;
A parameter setting method for a substrate inspection apparatus, wherein the obtained color range is set as a color condition used in substrate inspection. The substrate inspection apparatus according to claim 1, wherein the color space is a multidimensional color space including at least a saturation axis and a lightness axis for hues that tend to be included in the target image and hardly included in the excluded image. Parameter setting method. The parameter setting method for a substrate inspection apparatus according to claim 2, wherein the color condition includes a lower limit and an upper limit of saturation and a lower limit and an upper limit of brightness. The information processing apparatus is
Extracting a pixel region that satisfies the color condition from each of the target image and the excluded image, and creating a feature amount histogram for the feature amount of the pixel region,
Calculating a threshold for separating the feature amount of the target image and the feature amount of the excluded image based on the frequency distribution of the feature amount histogram;
The parameter setting method for a substrate inspection apparatus according to any one of claims 1 to 3, wherein the calculated threshold value is set as a determination condition used in substrate inspection. When an outlier exists in the frequency distribution of the target image in the feature amount histogram, a second color range is searched using the target image corresponding to the outlier and the plurality of excluded images,
5. The parameter setting method for a substrate inspection apparatus according to claim 4, wherein a set sum of the color range and the second color range is set as a color condition used in substrate inspection. When an outlier exists in the frequency distribution of the target image in the feature amount histogram, a second color range is searched using the target image corresponding to the outlier and the plurality of excluded images, and the outlier Search for a third color range using the non-target image and the plurality of excluded images;
5. The parameter setting method for a substrate inspection apparatus according to claim 4, wherein a set sum of the second color range and the third color range is set as a color condition used in substrate inspection. The target points mapped in the color space are divided into a plurality of groups based on the color distribution, and a color range is searched for each group.
The parameter setting method of the board | substrate inspection apparatus of any one of Claims 1-6 which set the set sum of the color range of all the groups as a color condition used by board | substrate inspection. A region satisfying a predetermined color condition is extracted from an image obtained by irradiating a component mounted on the board with light of multiple colors at different angles of incidence and imaging the reflected light, and the features of the extracted region Is an apparatus for automatically generating parameters used in a board inspection apparatus that inspects the mounting state of the component depending on whether or not a predetermined determination condition is satisfied,
Image acquisition means for acquiring a plurality of target images obtained by imaging components to be detected by inspection and a plurality of exclusion images obtained by imaging components to be excluded by inspection;
Mapping means for mapping the color of each pixel of the target image as a target point and mapping the color of each pixel of the exclusion image as a excluded point, respectively, in a color space;
Color range search means for obtaining a color range that divides the color space and that maximizes the difference between the number of target points included therein and the number of excluded points;
Color condition setting means for setting the obtained color range as a color condition used in substrate inspection;
A parameter setting device for a substrate inspection apparatus comprising: The substrate inspection apparatus according to claim 8, wherein the color space is a multidimensional color space including at least a saturation axis and a brightness axis for hues that tend to be included in the target image and hardly included in the excluded image. Parameter setting device. 10. The parameter setting device for a substrate inspection apparatus according to claim 9, wherein the color condition includes a lower limit and an upper limit of saturation and a lower limit and an upper limit of brightness. Area extracting means for extracting a pixel area that satisfies the color condition from each of the target image and the excluded image;
Feature amount histogram generating means for creating a feature amount histogram for the feature amount of the pixel area;
Threshold determination means for calculating a threshold for separating the feature amount of the target image and the feature amount of the excluded image based on the frequency distribution of the feature amount histogram;
Determination condition setting means for setting the calculated threshold as a determination condition used in substrate inspection;
The parameter setting device for a substrate inspection apparatus according to claim 8, further comprising: A storage unit that stores color conditions and determination conditions set by the parameter setting device according to claim 11;
A light projecting means for irradiating light of a plurality of colors at different incident angles on a mounting component on the substrate;
A region extracting means for extracting a region that satisfies the color condition from an image obtained by imaging the reflected light;
An inspection means for inspecting the mounting state of the component based on whether or not the feature value of the extracted region satisfies the determination condition;
A board inspection apparatus comprising:

Images