JP4470659B2 - Model registration method for parts inspection and inspection data creation apparatus using this method - Google Patents

Description

  The present invention relates to an inspection for determining whether a component mounting state is good or not using an image obtained by imaging a component mounting board, such as whether or not a component is missing and whether a mounting position and direction are appropriate. In particular, the present invention relates to a method for registering a model image for extracting a part to be inspected, and an inspection data creation apparatus to which the registration method is applied.

  In recent years, the applicant has developed an inspection apparatus that extracts components on a board by applying the invention described in Patent Document 1 below, and uses the extraction results to determine whether components are mounted or not. .

JP 2002-230549 A

  In the invention described in Patent Document 1, angle data (hereinafter referred to as “edge code”) indicating the direction of density change is obtained from a grayscale image to be processed, and this edge code distribution pattern is compared with a model pattern. Thus, an image region corresponding to the model is extracted. In a conventional inspection apparatus, a grayscale image of a model component mounted on a board is input prior to inspection, and a pattern image (hereinafter referred to as an “edge code image”) that associates each edge pixel on this image with its edge code. ”) And register it as a model pattern. Further, at the time of inspection, an edge code image is similarly created for the grayscale image to be processed, and the image pattern corresponding to the component is extracted by scanning the model pattern on the edge code image and collating it. .

  When registering a model pattern as described above, it is desirable to register the model pattern after verifying whether the created model pattern can correctly extract parts. For this reason, in the conventional inspection apparatus, the edge image of the reference board is displayed, and the extraction result of the component is displayed on the edge image so that the user can judge the accuracy of the component extraction. Yes.

  However, if a simple edge image is displayed, it is impossible to determine the direction of density change in the edge portion, and it may not be possible to correctly determine whether or not the result of component extraction is appropriate. For example, if noise is generated in the vicinity of a part and the noise and the outline of the part cannot be distinguished from each other, it is impossible to determine whether the part is correctly extracted or whether the extraction position is shifted. . In addition, when a part similar in shape to the extraction target part but having a different brightness relationship with respect to the substrate is in the vicinity of the extraction target part, it is about to distinguish the two parts from the outline on the edge image. Impossible.

  On the other hand, the edge code image can be displayed after gradation conversion, but it is difficult to grasp the contour shape of the component in such an image. Since the edge code is angle data in the range of 0 to 360 degrees, when this is replaced with the degree of shading, the edge pixel with a small edge code has almost the same density as the pixel that does not constitute the edge, and recognizes the edge. This is because it becomes difficult.

  The present invention has been made paying attention to the above-mentioned problems. By displaying an image that can easily grasp the direction of density change in various contour lines of a component to be extracted, the result of component extraction is appropriate. The purpose is to enable easy and accurate grasp of whether or not there is.

  The model registration method for component inspection according to the present invention extracts angle data indicating the direction of density change from a grayscale image of a model component mounted on a board, and corresponds the extracted angle data to a pixel at the extraction position. The attached pattern image is registered as a model pattern for component extraction. In this invention, the angle range that can be taken by the angle data is divided into a predetermined number of areas in advance, and individual colors are assigned to the respective areas, and the board on which the component corresponding to the model pattern is mounted is imaged. The angle data is extracted from the obtained grayscale image, a first step of generating a pattern image in which the extracted angle data is associated with the pixel at the extraction position, and the pattern image generated in the first step is the model. The second step of extracting the position and range of the image area corresponding to the model pattern on the pattern image by collating with the pattern, and the angle data in the pattern image generated in the first step as the angle data, respectively A third step of creating a color image replaced with the color of the corresponding region, and the position of the image region extracted in the second step And ranges are to be executed and a fourth step of generating and displaying the synthesized image on the color image.

In the above, the direction of density change is a direction in which the brightness changes with the edge pixel as a boundary, and can be considered to indicate a direction in which the dark portion changes to a bright portion, or the opposite direction. Specifically, as disclosed in Patent Document 1, it can be expressed by a combined vector of a density gradient vector in the horizontal direction and a density gradient vector in the vertical direction.
The angle data (corresponding to the edge code described above) indicating the direction of density change can be expressed by an angle with respect to a specific reference direction. In addition, the angle can be considered to be an arbitrary value within the range of 0 to 360 degrees. Note that it is desirable to extract this angle data only for pixels having an amount of change in density equal to or greater than a predetermined size (that is, edge pixels).

  The number of divisions of the region with respect to the angle range that the angle data can take can be set arbitrarily. However, considering that the components on the board are to be processed, it is divided into at least four areas so that the angles of 0 degrees, 90 degrees, 180 degrees, and 270 degrees are included in different areas. Is desirable. Arbitrary colors can be assigned to each area, but it is desirable to make the difference in color between the areas clear.

  According to the above method, after creating a pattern image to be a model pattern, the pattern image generated from the substrate image is collated with the model pattern to extract an image area corresponding to the component, and the suitability determination of the extraction result is made Display an image for. Since the image displayed here is a color image obtained by replacing the angle data in the pattern image to be processed with colors corresponding to the angle data, the user can display pixels represented by colors other than the background color. Can be recognized as edge pixels.

In the third step of creating a color image, the degree of color shading may be changed according to the amount of density change (edge strength) in each pixel. Preferably, the higher the edge strength, the darker the color.
According to such a method, the user can set a threshold value for extracting a pixel having a density change amount equal to or larger than a predetermined size while viewing a color shading display. As a result, it is possible to remove the color of the pixel in which the density change amount less than the predetermined size is caused by noise or the like and extract only the various contour lines of the component.

  In the above method, the model is based on the light / dark relationship between the component and the board body, and the light / dark distribution in the component (which reflects the difference in brightness among the component body, electrodes, solder, etc.). It is desirable to teach the user in what colors the various contour lines related to the parts corresponding to the are displayed. As a preferred method of this teaching, a similar color image can be created for the model pattern and displayed in parallel on the same screen as the color image on the substrate. In this way, the user can easily recognize the image area corresponding to the component by comparing the color image of the substrate to be processed with the color image of the model.

  It is desirable that the position and range of the image area combined with the color image of the substrate should not interfere with the color display of the angle data. For example, the inside of the extracted image region can be colored with a high brightness color that can transmit the display color of the angle data. Alternatively, the boundary line of the image area is displayed. When this boundary line is applied to the display color of the angle data, the boundary line can be set to be slightly shifted. By checking whether the image area displayed in this way corresponds to the part of the part determined from the color distribution of the color image, it is easily determined whether the part is correctly extracted from the model image. can do.

  An inspection data creation apparatus according to the present invention extracts an image input means for inputting a grayscale image necessary for creating inspection data, and angle data indicating the direction of density change from the image input by the image input means. A pattern image creating means for creating a pattern image in which the angle data is associated with the pixel at the extraction position, and storing a color associated with each predetermined number of areas divided from the angle range that the angle data can take. Color storage means, registration means for registering a pattern image created from the grayscale image as a model pattern for component extraction when a grayscale image of a model part is input from the image input means, and the image input means When a grayscale image of a board on which a component corresponding to the model pattern is mounted is input from the pattern image created from this grayscale image, Corresponding with the model pattern registered in the registration means, the corresponding area extracting means for extracting the position and range of the image area corresponding to the model pattern on the pattern image, and the area and color stored in the color storage means A color image creating means for creating a color image by replacing the angle data in the pattern image created from the grayscale image of the substrate with the color of the area corresponding to the angle, and the corresponding area extracting means Display control means for creating an image obtained by synthesizing the position and range of the image region extracted by the above on the color image and displaying the synthesized image.

  In the above inspection apparatus, the image input means can be configured by an interface circuit for inputting an image signal from the camera. Alternatively, an interface circuit for communication that receives transmission of image data to be processed from an external device, or a drive device for accessing a storage medium in which image data to be processed is stored can be used as the image input means. . In this apparatus, a monitor for displaying the composite image output by the display control means can be provided integrally with the cable connection or the apparatus.

  Each of the pattern image creating means, the corresponding area extracting means, the color image creating means, and the display control means can be configured by a computer in which a program for setting functions of the means is incorporated. Each means can be incorporated in the same computer. The color storage means and the registration means can be set in a memory in the computer or a memory externally attached to the computer.

  In the inspection data creation apparatus having the above configuration, first, a grayscale image of a model part is input, and a model pattern is created and registered from this input image. Next, the grayscale image of the board on which the component corresponding to the model is mounted is input, and the first to fourth steps described above are executed to display an image for determining the suitability of the component extraction result.

  The inspection data creation device can be configured by a general-purpose device such as a personal computer. Moreover, it can also introduce | transduce into the inspection apparatus used for the test | inspection of a component mounting state.

In the inspection data creation apparatus according to a preferred aspect, the registration unit registers a model color image in which the angle data in the model pattern is replaced with the color of the area corresponding to the angle, together with the model pattern. In addition, the display control means includes a means for inputting a rotation angle of the part relative to the part of the model as information indicating a mounting direction of the part corresponding to the model data on the board, and a model color image from the registration means. And a means for correcting the color of the read model color image based on the rotation angle , and outputting the model color image after the color correction together with the synthesized image.

  According to the above aspect, using the model color image created in advance from the model pattern, an image indicating the color that should appear on the outline of the component to be extracted is displayed together with the synthesized image. In this case, if the extraction target component is rotated with respect to the model component, the color of the model color image is corrected according to the rotation angle. Similar colors can be used. Therefore, the user can easily discriminate the outline of the component to be extracted by comparing the color distributions of the color image and the model color image on the substrate to be processed.

  According to the present invention, when a component extraction process is performed using a pattern image based on angle data indicating the direction of density change, a color corresponding to the size of the angle data in the image is distributed in the pattern image to be processed. Since it is converted into a color image and information indicating the component extraction result is combined with the color image and displayed, it can be easily and accurately determined whether the component extraction result based on the model image used is appropriate. can do.

FIG. 1 shows the configuration of a substrate inspection apparatus according to an embodiment of the present invention.
This board inspection apparatus is for processing an image obtained by imaging a board to be inspected to determine whether or not the mounting state of components on the board is appropriate. It comprises a control processing unit 5, an X-axis table unit 6, a Y-axis table unit 7, and the like.
In addition, 1T in the figure is a substrate to be inspected (hereinafter referred to as “inspected substrate 1T”). Further, 1S is a reference board with a good component mounting state, and is used for teaching prior to inspection.

  The Y-axis table unit 7 includes a conveyor 7A that supports the substrates 1S and 1T. The conveyor 7A is moved by a motor (not shown) to move the substrates 1S and 1T in the Y-axis direction (a direction orthogonal to the drawing sheet). ). The X-axis table unit 6 supports the imaging unit 3 and the light projecting unit 4 above the Y-axis table unit 7 and moves them in the X-axis direction (left-right direction in the figure).

  The light projecting unit 4 is composed of three annular light sources 8, 9, and 10 having different diameters. These light sources 8, 9, and 10 emit red (R), green (G), and blue (B) color lights, respectively, and the substrate is aligned with the center directly above the observation position. As viewed from 1S and 1T, they are arranged so as to be located in directions corresponding to different elevation angles.

  The imaging unit 3 includes a CCD camera for generating a color image, and is positioned so that its optical axis corresponds to the center of each light source 8, 9, and 10 and is along the vertical direction. As a result, the reflected light from the substrates 1S and 1T to be observed enters the imaging unit 3, is converted into color signals R, G, and B of the three primary colors and is input to the control processing unit 5.

  The control processing unit 5 includes a computer including a CPU as a control unit 11, and an image input unit 12, a memory 13, an imaging controller 14, an image processing unit 15, an illumination control unit 16, an XY table controller 17, an inspection unit 18, and a teaching table 19. , A data management unit 20, an input unit 21, a CRT display unit 22, a printer 23, a transmission / reception unit 24, an external memory device 25, and the like.

  The image input unit 12 includes an amplifier circuit that amplifies R, G, and B image signals from the imaging unit 3 and an A / D conversion circuit that converts these image signals into digital signals. The memory 13 stores digital gray image data for each color, binary image data obtained by processing these gray images, hue data, and the like. Further, when inspecting the component mounting state and creating data for the inspection, the memory 13 stores an edge image, an edge code image, and the like described later.

  The imaging controller 14 includes an interface for connecting the imaging unit 3 to the control unit 11. The imaging controller 14 drives the imaging unit 3 based on a command from the control unit 11 and adjusts the output level of each color light. Control. The illumination control unit 16 is for adjusting the light quantity of each light source of the light projecting unit 4. In this embodiment, the light amounts of the light sources 8, 9, and 10 are adjusted so that white illumination is performed by mixing red, green, and blue color lights.

  The XY table controller 17 includes an interface for connecting the X-axis table unit 6 and the Y-axis table unit 7 to the control unit 11, and based on commands from the control unit 11, the X-axis table unit 6 and the Y-axis table unit 7. Controls the movement movement.

  The teaching table 19 is a storage unit for storing board inspection data. In this inspection data, in addition to the setting position and size of the inspection area, model images of various parts, edge code images and code display image models to be described later are registered as inspection data for determining the component mounting state. To do.

  Prior to the inspection, the inspection data is taught by an attendant using an image obtained by imaging the reference board 1S or a model image of a part input in advance. These inspection data are collected as a determination file for each type of substrate. The data management unit 20 is a memory in which link information for associating a board type with a determination file is stored. After receiving the input of the board name of the board 1T to be inspected, the control unit 11 reads the determination file corresponding to the board 1T to be inspected based on the link information of the data management unit 20 and sets it in the memory 13. The image processing unit 15 and the inspection unit 18 execute processing based on the inspection information in the read determination file.

  At the time of inspection, the image processing unit 15 processes a color image with R, G, and B gradations stored in the memory 13 according to a procedure described later, and extracts an image area corresponding to the part. The inspection unit 18 compares the component extraction position with the reference mounting position of the component, thereby determining missing or misalignment of each component.

  The control unit 11 determines whether or not the substrate to be inspected 1T is a non-defective product by integrating the results of various determination processes in the inspection unit 18. The final determination result is output to the CRT display unit 22, the printer 23, or the transmission / reception unit 24.

  The input unit 21 is used to input various conditions for inspection, input of inspection information, and the like, and includes a keyboard and a mouse. The CRT display unit 22 (hereinafter simply referred to as “display unit 22”) receives supply of image data, inspection results, and the like from the control unit 11 and displays them on the display screen. The printer 23 also receives inspection results and the like from the control unit 11 and prints them out in a predetermined format.

  The transmission / reception unit 24 is for exchanging data with other devices. For example, for the substrate 1T to be inspected determined to be defective, the transmission / reception unit 24 transmits the identification information and the content of the defect to the subsequent correction device. Thus, the defective portion can be promptly corrected. The external memory device 25 is a device for reading / writing data from / to a storage medium such as a flexible disk, a CD-R, or a magneto-optical disk, and stores the inspection results and externally stores programs and setting data necessary for the inspection. Used to capture from.

  In the above configuration, the image processing unit 15 and the inspection unit 18 are configured by a dedicated processor in which a program for executing the above-described processes is incorporated. However, a dedicated processor is not necessarily provided, and the functions of the image processing unit 15 and the inspection unit 18 may be provided to the control unit 11.

  Moreover, when inspecting a substrate after soldering in the above-described substrate inspection apparatus, in addition to the above-described component mounting state inspection, a soldering state inspection can be performed. In this inspection, the arrangement of the light source of the light projecting unit 4 is used to image the specular reflected light from the soldered portion of the board, and based on the color distribution state on the obtained image, the solder inclination state Judge the quality.

Hereinafter, a method for inspecting the mounting state of the component and registering inspection data for the inspection will be described.
In this embodiment, the color image to be inspected is subjected to monochrome conversion, and an edge code is calculated for each edge pixel on the converted image. Next, a pattern image (edge code image) in which the edge code and the pixel are associated is created, and an edge code image (hereinafter referred to as “matching model”) of a model registered in advance in the edge code image is operated. The parts to be inspected are extracted by executing the matching process (hereinafter referred to as “edge code matching”). Here, when the component is extracted at the reference mounting position, it is determined that the component is properly mounted. On the other hand, if no part is extracted or if the extraction position deviates from the reference mounting position, it is determined that the part is missing or misaligned.

  A model of a grayscale image of a component registered as inspection data (hereinafter referred to as “model image”) is set by cutting out a small area including the component model from a color image of a board on which the component model is mounted. . The matching model used for the edge code matching is generated from an image obtained by monochrome conversion of the model image.

  Further, in this embodiment, when the user confirms whether the basic model image or the matching model is suitable for extracting a part, the edge code image created from the image of the reference board 1S is registered as a target. Edge code matching based on the matching model is executed, and the processing result is displayed on the display unit 22. On the display screen in this case, the edge code image or the matching model to be matched is displayed as a color image in which each edge code is replaced with a predetermined color. Furthermore, a colored display with a predetermined color (for example, yellow) is performed on an image region extracted as a component in the image display to be matched.

Here, a method of displaying the edge code image in color will be described.
The edge code is angle data indicating a direction in which the density changes with an edge pixel on the image as a boundary. Specifically, as shown in FIG. 2, a combined vector F of a vector fx based on the density change amount Δx in the x-axis direction and a vector fy based on the density change amount Δy in the y-axis direction is obtained, and the reference direction of the vector F An angle EC with respect to (in this embodiment, the positive direction of the x axis) is defined as an edge code. The vectors fx and fy in this embodiment are vectors indicating the direction from the darker side toward the brighter side. Therefore, the direction of the composite vector F also indicates the direction from the dark side to the bright side.

  The above edge code can be considered to take an arbitrary value within an angle range from 0 degrees to 360 degrees. Therefore, in this embodiment, as shown in FIG. 3, the angle ranges that can be taken by the edge code are divided into four regions R1, R2, R3, and R4 with each direction of 45 degrees, 135 degrees, 225 degrees, and 315 degrees as boundaries. The regions R1, R2, R3, and R4 are respectively associated with red, green, blue, and white colors. Then, when the edge code is calculated for each edge pixel on the image, the color of the region corresponding to the calculated edge code is set for each of the edge pixels, and a color image based on the setting is created. Hereinafter, this color image is referred to as a “code display image”.

  For example, a pixel whose edge code is 0 degree is red, and a pixel whose edge code is 90 degrees is green. Further, a pixel whose edge code is 180 degrees is blue, and a pixel whose edge code is 270 degrees is white. In this embodiment, the density gradient intensity (corresponding to the length of the vector F. This is hereinafter referred to as “edge intensity”) is obtained by using Δx and Δy for all the constituent pixels on the image. An edge code is calculated for a pixel whose edge strength is a predetermined value or more. Also, black is set as the background color for the pixels for which the edge code has not been calculated.

FIG. 4A is an example of an original color image obtained by imaging a part. In this figure, various colors are indicated by different fill patterns.
In the image of this example, 30 indicates a component, 31 indicates an electrode, and 32 indicates cream solder on the substrate. Further, the surrounding hatched portion 33 is a portion where the color of the substrate body appears. In the figure, an arrow set across the boundary position of each part schematically shows the direction of the vector F. In this example, the electrode 31 is brightest, and the components 30, the cream solder 32, and the substrate body 33 are darker in the following order.

  FIG. 4B is a code display image generated by replacing each edge code on the edge code image created from the image of FIG. 4A with the color of FIG. In FIG. 4 (2), contour lines of four types of colors of red, green, blue, and white are indicated by different types of lines. Note that the white background portion in the figure is a black background color in an actual image.

  In this embodiment, in addition to a model image and a matching model, a model of the code display image (hereinafter referred to as “display model”) is created and registered in the teaching table 19 for the component model. . 5 and 6 show a series of flows related to this registration process. The details of the procedure for registering the above three types of models will be described below along this flow.

  This procedure is started by inputting a color image of the reference board 1S on which a component model is mounted. In the first ST1 (ST is an abbreviation of a step. The same applies to the following), an operation for designating an area including the model of the part is received from the input image, and the area is set as a registration target area. Next, when a registration operation is performed by the user, ST2 becomes “YES” and the process proceeds to ST3, where a process of converting a color image in the registration target area into a monochrome image is executed.

  In ST4, the reference threshold value th registered as common data in the teaching table 19 is read out. The reference threshold value th is used to extract edge pixels from the registration target area, and is set to a value corresponding to a predetermined amount of edge strength or a numerical value in units of%.

  In the next ST5, after the density change amounts Δx and Δy are extracted for each pixel in the registration target region, the edge strength P is calculated. For calculating the edge strength P, the following equation (1) is used.

Next, in ST6, for the edge strength P, the maximum value _Pmax and the minimum value _Pmin in the registration target area are calculated. In further ST7, it is checked whether a unit of% is set for the reference threshold value th. If this determination is “YES”, the process proceeds to ST8, and the calculation according to the following equation (2) is executed using the maximum value P _max and minimum value P _{min of the} edge strength and the reference threshold value th, and edge pixel extraction is performed. A threshold value T is calculated. On the other hand, if the unit of% is not set for the reference threshold value th, ST7 becomes “NO” and the process proceeds to ST9, and the value of th is substituted for the threshold value T as it is.
T = P _min + (P _max −P _min ) × th (2)

  Below, paying attention to the pixels in the registration target region in order, the color data (R, G, B) of each pixel is determined by repeating the loop of ST10 to 22 for each pixel of interest. Note that R corresponds to a red gradation, G corresponds to a green gradation, and B corresponds to a blue gradation.

  In ST10, which is the starting step of the loop of ST10-22, the edge intensity P of the pixel of interest is compared with the threshold value T. If the edge strength P is smaller than the threshold value T, the process proceeds to ST21, and each gradation of the color data (R, G, B) is set to zero. As a result, black, which is the background color, is set for the pixel of interest.

  On the other hand, if the edge intensity P of the pixel of interest is greater than or equal to the threshold value T, ST10 is “YES” and the process proceeds to ST11 to calculate the edge code EC of the pixel of interest. Further, in the next ST12, the edge strength P is subjected to gradation conversion by the following equation (3), and the converted gradation is temporarily stored in the memory 13 as S. In the equation (3), A is an offset value for displaying each color with a visually recognizable intensity, and is set to A = 127, for example.

  After ST13, it is determined which of the four regions R1, R2, R3, R4 of FIG. 3 the edge code EC belongs to, and color data (R, G, B) is determined based on the determination result. . That is, when the edge code EC is smaller than 45 degrees or greater than 315 degrees (when ST13 is “NO” or ST16 “YES”), the process proceeds to ST17, and the color data (R, G, B) is stored. Set (S, 0, 0). As a result, the target pixel is set to red.

  When the edge code EC of the pixel of interest is 45 degrees or more and less than 135 degrees (when ST13 is “YES” and ST14 is “NO”), the process proceeds to ST18, and the color data (R, G, B) is obtained. Set to (0, S, 0). As a result, the target pixel is set to green.

  When the edge code EC of the pixel of interest is 135 degrees or more and less than 225 degrees (when ST14 is “YES” and ST15 is “NO”), the process proceeds to ST19, and the color data (R, G, B) is stored. Set to (0, 0, S). As a result, the target pixel is set to blue.

  When the edge code EC of the pixel of interest is 225 degrees or more and less than 315 degrees (when ST15 is “YES” and ST16 is “NO”), the process proceeds to ST20, and the color data (R, G, B) is obtained. Set (S, S, S). Thereby, the target pixel is set to white.

  In this way, edge pixels are extracted from the pixels in the registration target region, and the color of the region to which the edge code belongs and the gradation S corresponding to the magnitude of the edge strength P are set for each edge pixel. . When the process is completed for all the pixels in the registration target region, ST22 becomes “YES”, and thereafter, a process for registering various inspection data relating to the model of the part in the teaching table 19 is executed.

  In ST23, the threshold value T set in ST8 or ST9 is registered. Next, in ST24, an image in the registration target area is cut out from the color image input for the registration process, and this is registered as a model image. Furthermore, in ST25, an edge code image is created using the calculation result of the edge code of the pixel determined as the edge pixel by the threshold value T, and this is registered as a matching model. In this matching model image, pixels other than the edge pixels are set as blanks having no data value.

  Finally, in ST26, a color image based on color data (R, G, B) set for each pixel in the registration target area is registered as the display model.

  In this way, the threshold value T for edge extraction, the model image, the matching model, and the display model are registered for one part. Hereinafter, the combination of the threshold value T and the three types of image data is referred to as “model data”.

  In the board inspection apparatus of this embodiment, up to five pieces of the model data can be registered for one part. This is due to the presence of parts that have the same performance but differ in appearance due to differences in manufacturers. Since any part can be used as long as the performance is the same, model data of each part can be associated with a representative part among these parts.

  Furthermore, in the board inspection apparatus of this embodiment, processing for confirming whether or not the registered model data is appropriate is executed. FIG. 7 shows the flow of this confirmation process. This process is started from ST101.

In the board inspection apparatus at the time when this confirmation process is executed, information (part type, mounting position, mounting direction, etc.) of components mounted on the processing target board is registered based on CAD data or the like. Prior to the processing, when the board name of the reference board 1S on which the part to be checked is mounted is input, the part names of the parts on the board are read out and displayed in a list on the display unit 22.
The reference substrate 1S used for the confirmation process may be the same substrate as that used in the model registration process, but is preferably a different type of substrate. It is desirable to repeat the confirmation process using a plurality of types of reference substrates 1S.

  In ST101 of FIG. 7, the reference substrate 1T is carried in and imaged by a camera. Next, in ST102, when an operation for specifying a component to be confirmed, such as selecting a component name in the list display, is performed, the process proceeds to ST103, and the mounting direction is read from the registered information of the specified component. Further, in the next ST 104, model data corresponding to the designated part is read from the teaching table 19 and set in the memory 13. If a plurality of model data is registered for the designated part, all model data is read in ST104.

  In the next ST105, processing for correcting the color representing the edge code according to the mounting direction of the specified component is executed for the display model in the model data read in ST104.

  The model data of this embodiment is created by using an image of a part model positioned in the reference direction. The mounting direction of the component read in ST103 is shown as a rotation angle with respect to the component when facing the reference direction. If the orientation of the contour changes as the part rotates, the edge code also changes, so the color corresponding to the edge code may also change. In ST105, considering this point, correction is performed so that the edge code of the display model is the same as that of the actual part. Specifically, for each edge code EC in the matching model, a remainder [Mod ((EC + θ), 360)] obtained by dividing the value obtained by adding the rotation angle θ to the current value by 360 degrees is obtained as the correction value EC ′. The display model after correction is assigned the color corresponding to the correction value EC ′.

  Next, in ST106, the processing target image obtained in ST101 is subjected to monochrome conversion, and the converted image is subjected to the same processing as ST10 to 22 in FIG. 6 to thereby obtain an edge code image and code display. Create an image. Further, in the next ST107, the processing target image (color image) and the display model before the correction of ST105 are displayed in parallel on the display unit 22.

  When an operation for selecting one of the models is performed on this display, the process proceeds from ST108 to ST109, and edge code matching using a matching model corresponding to the selection is executed on the created edge code image. As a result, the image area having the highest similarity to the matching model on the edge code image is extracted as the area corresponding to the part.

  In ST110, the display screen of the display unit 22 is updated, and the image in which the image region extracted in ST109 is colored and displayed on the code display image created in ST106 and the display model corrected in ST105 are arranged in parallel. indicate. By comparing the two displays, the user checks whether the various contour lines in the extracted image area are colored in the same pattern as the display model, and determines whether the model is appropriate. If it is determined that the model is not appropriate, the user can perform a deletion operation. When this deletion operation is performed, ST111 becomes “YES” and the process proceeds to ST112, where processing for deleting the selected model data is performed.

  Thereafter, when an operation for selecting another model is performed, ST113 becomes “YES”, the process returns to ST109, and the same process as described above is performed using the newly selected model. On the other hand, if the end operation is performed without selecting a model, ST114 becomes “YES” and the confirmation process is ended.

  FIG. 8 shows an example of a display screen presented on the display unit 22 in the registration model confirmation process. FIG. 8A is a display screen before the selection operation in ST108 is performed. The model image of each model and the display model are associated with each other next to the window 40 that displays the color image to be processed. Is displayed. In the figure, reference numeral 41 denotes a model selection cursor.

  FIG. 8 (2) shows a display screen after model selection. On this screen, the display area of the selected model 1 is identified and displayed by the frame image 42. In addition, the display in the window 40 changes to a code display image, and a colored display is given to the image area 43 extracted by edge code matching. The coloring display is set to be slightly larger than the actually extracted region, and is colored with a color that can transmit the color of each contour line.

  In the above embodiment, the color on the display model side is changed in accordance with the mounting direction of the components. However, the present invention is not limited to this, and the color of the code display image in the window 40 is matched with the display model. May be corrected.

  In the example of FIG. 8, for the sake of simplicity, only the parts to be confirmed are displayed in the window 40. However, in the actual processing, the imaging area of the camera usually includes a plurality of parts. . For this reason, in the conventional method of displaying the extraction result of the part on the edge image, when there is a part with a similar shape in the vicinity of the part to be confirmed, or when there is noise similar to a part of the part, It becomes difficult to determine whether a part has been correctly extracted.

  On the other hand, in this embodiment, the color pattern in the region 43 extracted as a part can be compared with the color pattern in the selected model, and it can be confirmed whether or not the part is correctly extracted by edge code matching. The accuracy and efficiency of the confirmation work can be greatly improved.

  In the above confirmation processing, an edge code image or a code display image created from a monochrome image is used. However, when the accuracy of edge code matching is poor, any gradation of R, G, or B is used. It is also possible to re-create the edge code image or code display image using the gray image obtained by the above or the gray image subjected to brightness correction, and perform edge code matching again.

1 is a block diagram of a substrate inspection apparatus to which the present invention is applied. It is explanatory drawing which shows the concept of an edge code. It is explanatory drawing which shows the example which divided the area | region which the angle code | cord | chord can take, and allocated the color. It is explanatory drawing which contrasts and shows the grayscale image of components, and the code display image produced | generated from this image. It is a flowchart which shows the procedure of a model registration process. FIG. 6 is a flowchart continued from FIG. 5. It is a flowchart which shows the procedure of the confirmation process of a registration model. It is explanatory drawing which shows the example of a display in the confirmation process of FIG.

Explanation of symbols

1T reference board 3 imaging unit 12 image input unit 11 control unit 15 image processing unit 19 teaching table 22 CRT display unit

Claims (3)

Extract angle data indicating the direction of density change from the grayscale image of the model component mounted on the board, and register the pattern image associating the extracted angle data with the pixel at the extraction position as a model pattern for component extraction In the way to
The angle range that the angle data can take in advance is divided into a predetermined number of areas, and individual colors are assigned to each area,
First, the angle data is extracted from a grayscale image obtained by imaging a board on which a component corresponding to the model pattern is mounted, and a pattern image in which the extracted angle data is associated with a pixel at the extraction position is generated. Steps,
A second step of collating the pattern image generated in the first step with the model pattern and extracting a position and a range of an image region corresponding to the model pattern on the pattern image;
A third step of creating a color image by replacing the angle data in the pattern image generated in the first step with the color of the region corresponding to the angle data;
A model registration method for component inspection, comprising: a fourth step of creating and displaying an image obtained by combining the position and range of the image region extracted in the second step on the color image. An apparatus for creating inspection data for inspecting a component mounting board,
An image input means for inputting a grayscale image necessary for creating inspection data;
Pattern image creation means for extracting angle data indicating the direction of density change from the image input by the image input means, and creating a pattern image in which the extracted angle data is associated with the pixel at the extraction position;
Color storage means for storing individual colors in association with each of a predetermined number of areas divided from an angle range that can be taken by the angle data;
A registration unit that registers a pattern image created from the grayscale image as a model pattern for component extraction when a grayscale image of the model component is input from the image input unit;
When a grayscale image of a substrate on which a component corresponding to the model pattern is mounted from the image input means, a pattern image created from the grayscale image is collated with a model pattern registered in the registration means, A corresponding region extracting means for extracting a position and a range of an image region corresponding to the model pattern on the pattern image;
Based on the correspondence between the area and color stored in the color storage means, a color image is created by replacing the angle data in the pattern image created from the grayscale image of the substrate with the color of the area corresponding to that angle. Color image creation means to
An inspection data creating apparatus comprising: display control means for creating an image obtained by synthesizing the position and range of the image area extracted by the corresponding area extracting means on the color image and outputting the synthesized image for display. A model color image in which the angle data in the model pattern is replaced with the color of the area corresponding to the angle is registered in the registration unit together with the model pattern,
The display control means reads out a model color image from the registration means and means for inputting the rotation angle of the part relative to the model part as information indicating the mounting direction of the part corresponding to the model data on the substrate. 3. The inspection data creation according to claim 2, further comprising: means for correcting the color of the read model color image based on the rotation angle , and outputting the model color image after the color correction together with the synthesized image. apparatus.

Images