JP5045591B2 - Method for creating area setting data for inspection area and board appearance inspection apparatus - Google Patents

Description

  The present invention belongs to the technical field of automatically inspecting the mounting state of each component on a board using an image generated by imaging the component mounting board by an imaging means. In particular, the present invention relates to a method of creating inspection data used for this type of automatic inspection using a computer and a substrate appearance inspection to which this method is applied.

  In the substrate appearance inspection, various inspections such as presence / absence of components, appropriateness of position and orientation, and appropriateness of soldering state are performed on each component of the substrate to be inspected. In order to automatically perform this type of inspection, inspection data such as the position of each part, inspection area setting data, parameters (for example, binarization threshold) used for detection of the part to be inspected, thresholds for determination, etc. Must be registered.

  As a conventional method for automatically creating the inspection data, a method using CAD data or data used in a manufacturing process (mount data, Gerber data, etc.) is known. For example, the following Patent Document 1 describes the position of a camera field of view and a check area in the field of view using mount data used to mount a chip on a substrate and chip data representing the external shape of each chip. It describes that the position of (inspection area) is obtained.

  Patent Document 2 obtains information representing the shape and position of an element solder printing portion (printed portion of cream solder) formed by printing from mask opening data (gerber data), and represents the geometrical information represented by this information. It describes that ranges are associated with component mounting data and component data libraries of various components.

Japanese Patent No. 2940213 JP 2007-218925 A

  In recent years, there have been an increasing number of cases where PCB mounting is outsourced to EMS (Electronics Manufacture Services) contractors. Such subcontractors are provided with highly confidential information such as CAD data. In many cases, it is difficult to create inspection data.

  The present invention has been made paying attention to the above points, and it is an object of the present invention to make it possible to automatically create area setting data for an inspection area of each component without using board design data such as CAD data.

  The method of creating the area setting data of the inspection area according to the present invention is to create the area setting data used for setting the inspection area for each component in the image in the automatic appearance inspection for the component mounting board by a computer, The following first to fifth steps are performed.

  In the first step, the standard arrangement pattern of the land area and the standard area setting data corresponding to the component electrodes are registered in the memory of the computer for a plurality of component types. In the second step, the land area on the board, the silk print pattern, and the wiring pattern are individually detected from the model board image before or after component mounting, and three types of binary images representing each detection result are generated. To do.

  In the third step, the land areas are grouped on the basis of the correspondence relationship of the three types of binary images on condition that the group of land areas having a relationship involving silk print patterns or wiring patterns is not classified into the same group. .

  In the fourth step, after the above grouping is completed, the process of collating the binary image of the land area with the standard arrangement pattern of the land area for each component type is made to associate the land regions in the same group with the same component type. By executing this as a condition, the component type, the mounting range, and the component orientation of each component mounted on the board are specified.

  In the fifth step, by applying the standard region setting data of the component type specified based on the orientation of the component specified in the specified mounting range to the component specified in the fourth step, the component The inspection area to be set is specified, and the area setting data for setting the inspection area is registered in the memory of the computer.

  In the second step, when processing an image of a model board before component mounting, it is desirable to detect the entire range of lands as a land area. On the other hand, when processing an image of a model board after component mounting, a land soldered range is detected as a land area.

  Even if the board design data is not used, if the binary image of the land area created from the image of the model board can be associated with the standard arrangement pattern of the land area for various component types, the component type of each component on the substrate, It is possible to identify the mounting range and the component orientation, and to determine the inspection region by applying standard region setting data for each component type to the identification result. However, if the binary image of the land area is simply compared with the standard arrangement pattern, a plurality of land areas that should be associated with the same part can be associated with different parts or a plurality of parts that should be associated with different parts. There is a possibility that an error occurs in the correspondence between the land area and the part, for example, the land area is associated with the same part.

  Therefore, in the method according to the present invention, land groups corresponding to the same part are grouped by grouping the land areas on the condition that a set of land areas having a relationship involving silk printing patterns or wiring patterns is not classified into the same group. It becomes possible to set a group in which only areas are collected (a plurality of groups may be set for one part). In addition, after grouping, the land area and the part can be compared with each other by collating with the standard area setting data of various part types on condition that the land area in the same group is associated with the same part type. Correspondence errors can be prevented, and the component type, mounting range, and component orientation of each component can be accurately identified.

  In a preferred aspect of the above method, in the third step, one group is set by land areas arranged in a state surrounded by a silk print pattern. Since the region surrounded by the silk print pattern represents the mounting range of one component, it is possible to easily specify a group of land regions corresponding to the same component.

  In another preferred embodiment, in the third step, a plurality of land regions whose areas and shapes match with each other are arranged along a certain direction without any silk printing pattern and wiring pattern interposed therebetween. Classify into the same group. According to this method, a set of land areas corresponding to a plurality of leads arranged along one side of the component package can be included in the same group.

  In another preferable region, in the third step, a plurality of land regions having the same area and shape and arranged without any silk printing pattern and wiring pattern are classified into the same group. According to this method, a set of land regions corresponding to components having a relatively small number of electrodes, such as chip components and transistors, can be included in the same group.

  Next, in another preferred aspect, in the fourth step, the standard pattern of one component type is associated with the distribution pattern of one or more groups of land areas, and the number of land areas and the degree of arrangement match are obtained. In this way, even if the land areas corresponding to one part are all classified into the same group or divided into a plurality of groups at the stage of grouping in the third step, The degree of coincidence when these groups are correctly associated with the corresponding part type can be made higher than in other cases, and the parts can be identified with high accuracy.

  The area setting data creation method includes an imaging unit for imaging a substrate, and an area of an inspection region set for each component in an inspection target image generated by imaging the substrate to be inspected by the imaging unit. Area setting data creating means for creating setting data, a memory for storing the created area setting data, an inspection area is set in the inspection target image based on the area setting data registered in the memory, and each inspection area The present invention can be applied to a board appearance inspection apparatus that includes an inspection execution unit that determines the mounting state of each component on the substrate by processing an image in the image, and an output unit that outputs a determination result by the inspection execution unit.

  In the above-described inspection apparatus, for the plurality of component types, standard arrangement patterns of land areas corresponding to the component electrodes and standard inspection data including standard area setting data of the inspection area are registered. The area setting data creating means includes the following binary image generating means, grouping means, component specifying means, and registration means.

  The binary image generating means individually detects a land area, a silk print pattern, and a wiring pattern on the board from an image of the model board before or after component mounting, and three types of binary images representing each detection result. Is generated. The grouping means groups the land areas based on the correspondence relationship of the three types of binary images on the condition that the set of land areas having a relationship involving the silk print pattern or the wiring pattern is not classified into the same group. After the grouping is completed, the component specifying means performs a process of collating the binary image of the land region with the standard arrangement pattern of the land region for each component type, and associating the land region in the same group with the same component type. By executing as a condition, the component type, mounting range, and component orientation of each component mounted on the board are specified. The registration unit applies the standard region setting data of the component type specified based on the direction of the component specified in the specified mounting range to each component specified by the component specifying unit, thereby The inspection area to be set is specified, and the area setting data for setting the inspection area is registered in the memory.

  According to the region setting data creation method and the board appearance inspection apparatus, the correspondence between the three types of binary images generated from the model board image and the standard arrangement pattern of the land area for each component type are used. It is possible to correctly specify the component type, mounting range, and component orientation of each component from the distribution pattern of the land area in the image. Therefore, it is possible to determine the inspection region by correctly associating each region with the standard region setting data of the corresponding component type without using the board design data.

(1) Apparatus Configuration FIG. 1 shows an electrical configuration of a substrate visual inspection apparatus to which the present invention is applied.
This board appearance inspection apparatus (hereinafter simply referred to as “inspection apparatus”) is for inspecting the suitability of the position and orientation of each component, soldering state, etc., on a printed circuit board after soldering. 1, an illumination device 2, an X stage unit 3, a Y stage unit 4, and a control processing device 50 including a computer control unit 5. In addition, although not shown in FIG. 1, the inspection apparatus is provided with a substrate support table for supporting a substrate to be inspected, a substrate carry-in / out mechanism, and the like.

 The camera 1 and the illumination device 2 constitute an optical system called “color highlight system”. The camera 1 is of a type that generates a color still image, and is arranged above the substrate with the light receiving surface facing directly below. The illuminating device 2 irradiates red, green, and blue color lights from different directions surrounding the optical axis of the camera 1 with different incident angles.

  The X stage unit 3 supports the camera 1 and the illumination device 2 above the substrate support table, and the Y stage unit 4 supports the substrate support table. In any of the stage portions 3 and 4, the support target can be moved along one axis. The direction of movement by one stage unit is orthogonal to the direction of movement by the other stage unit.

  The control processing device 50 controls the operation of each of the stage units 3 and 4 and the camera 1 and the illumination device 2 to generate a color image for inspection, and executes various inspections. 6, an image input unit 7, an imaging control unit 8, an illumination control unit 9, an X stage drive unit 10, a Y stage drive unit 11, an input unit 12, a display unit 13, and a communication interface 14.

  The image input unit 7 includes an interface circuit that receives R, G, and B image signals output from the camera 1 and an A / D conversion circuit that digitally converts these image signals. The imaging control unit 8 controls the imaging timing of the camera 1, and the illumination control unit 9 controls the light amount, emission color, lighting timing, and the like of the lighting device 2.

  The input unit 12 is for performing a setting operation at the time of teaching, and is configured by a dedicated operation button, a mouse, or a keyboard. The display unit 13 is used to display an inspection image, an inspection result, and the like, and includes a liquid crystal panel or the like. The communication interface 14 is used for the purpose of transmitting the inspection result to an external device.

  The memory 6 is a large-capacity storage device such as a hard disk, and stores an inspection data file 61, a land pattern registration unit 62, a standard inspection data registration unit 63, and the like, in addition to programs related to control and inspection.

  In the inspection data file 61, in addition to setting data (data indicating the position and size of various areas) for setting the inspection area for each part in the image, the binarization for detecting the color of the inspected part is performed. A threshold value and a threshold value for determination for determining the suitability of the feature amount obtained by measuring the inspection site are registered.

  In the land pattern registration unit 62, standard arrangement patterns of lands of a plurality of component types (hereinafter referred to as “standard land patterns”) are registered, and in the standard inspection data registration unit 63, standard inspection of each component type is performed. Data is registered. These databases 62 and 63 are used to create inspection data to be registered in the inspection data file 61, and any registered information is associated with component type identification information (component type name).

  Based on the program in the memory 6 and the inspection data file 61, the control unit 5 uses the image input from the camera 1 to execute image processing and determination processing for each component, and finally each inspection result. In summary, it is determined whether the substrate is a non-defective substrate or a defective substrate. The results of each inspection are collected for each substrate and displayed on the display unit 13, and are also output to an external device (not shown) via the communication interface 14.

(2) Processing for creating inspection data In the inspection apparatus, prior to inspection, an image (hereinafter referred to as “model image”) of a model of a substrate to be inspected (hereinafter referred to as “model substrate”) and land pattern registration. Using the registration information of the unit 62 and the standard inspection data registration unit 63, the inspection data of the substrate to be inspected is created and registered in the inspection data file 61. A series of processing relating to the creation and registration of inspection data is also executed by the control unit 5 that operates according to a program stored in the memory 6.

  FIG. 2 is a schematic flowchart regarding the automatic creation process of inspection data. Hereinafter, a series of processes executed by the control unit 5 will be described along the steps (S1 to S11) in FIG.

(A) Imaging of model substrate (step S1)
In this embodiment, a single model board is imaged before and after component soldering, component mounting, and soldering steps before and after each step. A model image of the board and a model image of the board after component mounting are generated. Hereinafter, the former is referred to as “pre-mounting model image”, and the latter is referred to as “post-mounting model image”.
Of these two model images, the pre-mounting model image plays a major role in the inspection data creation process.

  FIG. 3 shows an example of a pre-mounting model image. In this figure, the color of the land is represented by a paint pattern, the silk print pattern is represented by a thick black line, and the wiring pattern is represented by a line with a paint pattern or a single line. The actual land is close to ocher and the silk print pattern is close to white. The wiring pattern has a color (dark green) close to the color of the resist on the substrate (green).

  The formation position of each land is determined according to the position and orientation of the component to be mounted. Silk printing patterns are roughly classified into a frame-like pattern surrounding a mounting range of one component and a linear pattern representing a boundary line between adjacent components. The wiring pattern represented by the paint pattern forms a main signal path, and the single-line wiring pattern connects adjacent lands.

  Note that regions P and Q in FIG. 3 correspond to the illustration of FIG. 13 described later. The region S corresponds to the illustrations of FIGS. 16 and 18, and the region T corresponds to the illustrations of FIGS. 11 and 19.

  In this embodiment, lands, silk printing patterns, and wiring patterns are individually detected from the pre-mounting model image, and a binary image indicating each detection result is generated. Hereinafter, each binary image is referred to as “land pattern image”, “silk pattern image”, and “wiring pattern image”.

(B) Land pattern image generation and land recognition (step S2)
In step S2, the pre-mounting model image is binarized so that the pixels having the land color are “1” and the other color pixels are “0”, and noise removal processing is performed on the binarized image. As a result, a land pattern image as shown in FIG. 4 is generated.

  Further, in step S2, by performing a labeling process on the pixel whose value of the land pattern image is “1”, each land included in the land pattern image is individually recognized, and an identification number is set for each. . Also, the coordinates of the representative points of each land (for example, a circumscribed rectangle is set on the land and each vertex thereof is used as the representative point), and these coordinates and the identification number are associated with each other in the working memory (RAM). Save to. These stored data are used thereafter in processing for specifying a land to be processed or searching for a land to be included in a group (FIGS. 10 and 12).

  Note that the labeling process is to detect pixel groups having data of interest as individual areas by setting the same label to pixels having the same and continuous data.

(C) Generation of silk pattern image / wiring pattern image (steps S3 and S4)
Next, in step S3, the pre-mounting model image is binarized so that the pixels having the color of the silk print pattern become “1” and the pixels of the other colors become “0”, and noise is removed. And a process of complementing the linear pattern, etc., to generate a silk pattern image as shown in FIG.

  Next, in step S4, the pre-mounting model image is binarized so that the pixels having the color of the wiring pattern become “1” and the pixels of the other colors become “0”, and noise removal and straight lines are performed. Perform processing to complement the pattern. Note that in the noise removal processing in this case, in consideration of the fact that the color of the wiring pattern is close to the color of the resist and a part of the resist may be detected as noise, the region detected by binarization is used. Of these, those that are not linear patterns are removed as noise.

(D) Land grouping processing (steps S5 to S8)
In these steps, the lands are grouped based on the correspondence relationship with the silk pattern image and the wiring pattern image based on the land pattern image generated in step S2. In this series of grouping, it is a condition that lands sandwiched between silk print patterns or connected by wiring patterns are not included in the same group. Since the silk print pattern represents the boundary of the mounting range of each component, the silk print pattern cannot be interposed between lands corresponding to the same component, and there is no land corresponding to the same component. This is because there is no possibility of a short circuit via the wiring pattern.

  Furthermore, in this embodiment, in the course of the grouping process, the component type corresponding to the set group (hereinafter referred to as “land group”) is estimated, and each land group is determined as a connector / transistor group based on the estimation result. , IC group, and general land group. The information of each land group is obtained by associating the group identification number with information indicating the type of the group and the land identification number of the group component.

FIG. 6 shows a result of performing the grouping process on the land pattern image of FIG.
In this figure, the relationship between each group and the land is represented by using rectangular frames and paint patterns having different line types.

  According to the example of FIG. 6, the IC group includes two or more arrays in which narrow lands are arranged along the width direction. There are many types of connectors / transistor groups and general land groups that include a plurality of lands having the same shape, but there are also groups formed of only one land.

  Hereinafter, steps S5 to S8 in FIG. 2 will be described using detailed flowcharts and explanatory diagrams, respectively.

(D) -1 Grouping of lands based on silk inclusion area (step S5)
This process uses a silk pattern image to detect an independent area surrounded by a silk print pattern (referred to as “silk inclusion area”), and sets a land group for each silk inclusion area. FIG. 7 shows a detailed procedure, and FIG. 8 shows a specific example.

  Referring to FIGS. 7 and 8, first, a labeling process is executed on a background pixel of a silk pattern image (a pixel having no color of a silk print pattern, that is, a pixel having a value of “0”). (Step 501). Thereby, the area | region (R1-R13 of FIG. 8) in the state parted from the circumference by the silk printing pattern is detected. In addition, since the background pixels at locations other than the regions R1 to R13 are continuous at any position, a set of these background pixels is detected as one region R14.

  In this embodiment, using the above processing results, the region R14 having the largest area among the regions detected by the labeling processing is excluded, and the remaining regions R1 to R13 are specified as silk inclusion regions (step 502). FIG. 8 (2)). Thereafter, the identified silk inclusion areas R1 to R13 are associated with the land pattern image (step 503), and attention is paid to the silk inclusion areas R1 to R13 in order (step 504), and the lands included in the attention area are detected. One land group is set (steps 505 and 506). A group having two or less lands is defined as a general land group, and a group having three or more lands is defined as a connector / transistor group (steps 507, 508, and 509).

  According to the example of FIG. 8C, each of the silk inclusion regions R1 to R13 excluding the region R6 includes lands, and land groups are formed for these regions. Is done. Of these, regions R1, R3, and R12 are set as connector / transistor groups, and the remaining regions are set as general land groups.

(D) -2 Land distribution outside the silk inclusion region (step S6)
In this process, for the lands located outside the silk inclusion region, each land is connected to a land for connecting an electrode (lead) for a lead component (hereinafter referred to as “lead land”), and the lead component. It is classified as a land for connecting electrodes of other parts (hereinafter referred to as “general land”). FIG. 9 shows the detailed procedure of this distribution process.

  Referring to FIG. 9, first, attention is paid to one of unclassified lands (step 601), and the area of this land and the length of each side of the circumscribed rectangle are measured (step 602). Next, the ratio of the length of the short side to the long side of the circumscribed rectangle (short side length / long side length) is obtained, and if the value of this ratio is equal to or greater than a predetermined threshold value, If the ratio is smaller than the threshold value, the target land is identified as a read land (steps 603, 604, and 605). Further, each measurement value obtained in step 602 is temporarily stored in association with the identification number of the target land (step 606). In the following, the same processing is executed by paying attention to the unclassified lands in order (step 607).

(D) -3 Grouping of lead lands (step S7)
In this process, grouping is performed on the land specified as the lead land by the above-described distribution process. FIG. 10 shows the detailed procedure.

  Referring to FIG. 10, first, focusing on one of unclassified lead lands, a new group having this land as a constituent element is set (step 701).

  Next, along the width direction of the target land (the direction of the short side of the rectangle circumscribing the land), a land whose size and shape approximate to the land is searched (step 702). Specifically, a land whose difference between the measured value of each side or area of the circumscribed rectangle and the target land is within a predetermined allowable value and whose positional deviation in the length direction is within a predetermined allowable range is set as a detection target. . However, the search range is limited to a predetermined distance from the target land, and a search across the silk print pattern and the wiring pattern cannot be executed.

  If a land satisfying the above conditions is found, the detected land is added to the group (steps 703 and 704), and the search is continued. In the following, if a land to be detected is found, the land is added to the group being processed.

  In the above search, the first time, the search is performed in two directions that are reversed with the target land in between. If one of these lands is detected, the search is limited to that direction thereafter. Execute. However, if a land satisfying a condition in any direction is found in the first search, each direction is also set as a search target in the following.

  When the loop of steps 701 to 704 ends, the type of group is determined according to the number of lands included in the group set by this processing. Specifically, a group having two or more lands is set as an IC group, and a group having one land is set as a connector / transistor group (steps 705, 706, and 707).

  In the following, each of the read lands is classified into a predetermined number of groups by executing the above-described processing for the read lands for which no group is defined (701 to 708). Thereafter, the IC group is formed for each component by executing group integration processing (step 709) for the IC group.

Here, the integration processing will be described with reference to FIG.
In the example of FIG. 11, the four IC groups G1 to G4 set in the region T of FIG. 3 are integrated. Here, focusing on one of the set IC groups (G1), a group having a similar land arrangement that is opposed to this group G1 is searched ((1) (2) in FIG. 11). ). Specifically, a group in which the number of lands and the number of lands is the same and the size and shape of each land are approximated is searched for along the direction orthogonal to the direction of land arrangement of the group of interest. Although not shown in FIG. 11, the search in this case is also performed in two directions that are reversed across the array of the group of interest, and the search range is also limited.

  When an IC group G2 satisfying the conditions is found by the above processing, a group whose arrangement direction is rotated by 90 degrees with respect to the groups G1 and G2 is further searched (FIG. 11 (3)). In the search in this case, the search target direction is limited to the direction from one group to the other group.

  In this way, when the groups G3 and G4 whose arrangement direction is orthogonal to the groups G1 and G2 are found, the detected groups G2, G3 and G4 are integrated into the group of interest (FIG. 12 (4)).

  In the lead component mounting range, a silk print pattern may be formed between the arrangement of the leads or within the range surrounded by the lands (the pattern in the region T in FIG. 3 and the region R6 in FIG. 8). (Refer to the corresponding silk print pattern.) Therefore, when integrating the IC groups, the search across the silk print patterns is allowed exceptionally.

  In the example of FIG. 11, four unclassified lands 101 to 104 are included in a range surrounded by the integrated arrays. These lands 101 to 104 are for connecting electrodes on the lower surface of the package of the components. When the components are mounted, the lands 101 to 104 are hidden and are not subjected to appearance inspection. In view of this point, in this embodiment, when an unclassified land exists in the range surrounded by the integrated group, the land is invalidated and excluded from the processing target thereafter.

  According to the example of FIG. 11, with respect to a component (QFP) having a structure in which leads are arranged on all four sides of the package, land groups corresponding to the arrangement of leads on each side can be integrated into one group. Although not shown in the figure, the lead (SOP) having a configuration in which leads are arranged only on two opposite sides of the package is also used in the same manner as in (1) and (2) of FIG. Land groups corresponding to the array can be identified and combined into one group.

(D) -4 General land grouping (step S8)
This grouping is executed for the lands assigned to the general lands in step S6. FIG. 12 shows the detailed procedure.

  Referring to FIG. 12, in this process as well, a new group having the target land as a constituent element is set by focusing on one of the unclassified general lands (step 801). Next, the periphery of the target land is searched to detect a general land adjacent to the target land (step 802). The general land detected here is positioned as a candidate land that may be included in the same group as the target land.

  After this, whether the candidate land satisfies these conditions as a detection condition that the size and shape approximate to the target land and that there is no silk print pattern or wiring pattern between it and the target land. Please check. If a candidate satisfying these conditions is found, the candidate land is added to the group being processed (steps 803 to 806).

  In FIG. 12, in order to avoid complication, the processing procedure related to the search is simply expressed as in steps 802 to 806. In practice, in addition to the process of checking the candidates detected in step 802, these procedures are also performed. It is checked whether or not there is a land satisfying the above search condition even for a land newly added to the group from the candidates.

  In this way, when a predetermined number of lands are added to the group being processed to form one group, a group having two or less lands is set as a general land group as in the example of FIG. A group having three or more lands is set as a connector / transistor group (steps 807, 808, and 809).

  Thereafter, the above processing is repeated until there are no lands for which no group is set (steps 801 to 810), thereby classifying the general lands into a predetermined number of general land groups and connector / transistor groups.

  FIGS. 13A and 13B show an example in which a land belonging to the same group as this land is detected by paying attention to one of the lands included in the regions P and Q of FIG. Each example shows a schematic diagram in which three types of pattern images are superimposed on the left side, and the group setting result for the land pattern image is shown on the right side.

FIG. 13A is an example in which group setting is performed by paying attention to the land 200 in the region P.
When attention is paid to the land 200, the land 201 on the left and the land 202 on the lower left are detected as candidates, but the land 201 is connected to the target land 200 by the wiring pattern 205. Does not meet. On the other hand, since the land 202 is arranged next to the target land 200 without interposing a wiring pattern or a silk print pattern, the condition for entering the group is satisfied.

  Next, as the land 202 is put into the group, the land 203 on the left side becomes a new candidate. Since this land 203 is connected to the land 202 by the wiring pattern 206, the condition for entering the group is satisfied. Absent. Therefore, in this case, a general land group including the land 200 and the land 203 is set.

  FIG. 13 (2) is a group setting by paying attention to the lands 210 in the region Q. For this land 210, the adjacent land 211 on the left and the upper and lower lands 212 and 214 are detected as candidates, but the wiring patterns 215 and the silk print patterns 216 and 217 are interposed between the lands 212 and 214. Therefore, the conditions for entering the group are not met. On the other hand, the land 211 is arranged next to the land of interest 210 without interposing a wiring pattern or silk print pattern, and therefore satisfies the conditions for entering the group.

  Next, as the land 211 is added to the group, the land 213 on the land 211 becomes a new candidate. Since the silk printing pattern 216 is interposed between the land 213 and the land 211, the land 213 is included in the group. The condition is not met. Therefore, in this case, a general land group including the land 210 and the land 211 is set.

(E) Component identification processing (step S9)
This process is performed by performing a matching process using the standard land pattern of each component type on the land group set by the above various groupings, thereby changing the component type, mounting range, and component direction of each component on the board. It identifies and associates lands. Specifically, after the processing of FIG. 14 is executed for the connector / transistor group, the processing shown in FIG. 15 is executed for each group for the IC group and the general land group.

(E) -1 Component Identification Processing for Connector / Transistor Group First, component identification processing for a connector / transistor group will be described with reference to FIG. First, paying attention to one unprocessed connector / transistor group (step 901), one standard land pattern of a connector or transistor is read from the land pattern registration unit 62 (step 902).

  Next, the degree of coincidence between the read standard land pattern and the land pattern of the group of interest is calculated (step 903). If the degree of matching is equal to or less than a predetermined threshold, the degree of matching is calculated again for the range including the group adjacent to the group of interest. The adjacent group in this case is limited to a group in which the corresponding component is not specified, but is not limited to the connector / land group, and may be a general land group or an IC group.

  In both steps 903 and 905, the number of lands in the group to be collated is first compared with the number of lands in the standard land pattern, and the degree of coincidence is set to 0 if they do not match. Thereby, the matching by the standard pattern which does not correspond to a group at all can be substantially skipped.

  If the number of lands matches, the standard land pattern is associated with the range corresponding to the group to be matched in the land pattern image, and the land in the land pattern image matches the land on the standard land pattern side. Count the number of pixels. This calculation is not limited to one time, and the count process is repeated while changing the position and orientation of the standard land pattern, and the maximum value among them is compared with the threshold value.

  In the above processing, when the matching degree by only the target group or the matching degree including the adjacent group exceeds a predetermined threshold value, step 907 is executed. In this step, the degree of coincidence exceeding the threshold value is temporarily stored in association with the standard land pattern identification code. In addition, information indicating the position and rotation angle of the standard land pattern when the degree of coincidence is calculated (for example, the coordinates of each land of the standard land pattern and the rotation angle with respect to the reference posture) is temporarily stored (step 907). .

  Hereinafter, the standard land patterns corresponding to the various connectors and the standard land patterns corresponding to the various transistors are read in order, and the degree of coincidence with each is calculated (steps 902 to 908). When a series of processing is completed, the maximum value of the matching levels exceeding the threshold is extracted, and the part type of the part corresponding to the target group is specified based on the stored data when the maximum matching level is obtained. (Steps 909 and 910) Further, the region corresponding to this part and the direction of the part are specified (step 911).

  Specifically, in step 910, the component type corresponding to the standard pattern when the maximum degree of coincidence is obtained is identified as corresponding to the group of interest. In step 911, a rectangular area including the distribution range of each land associated with the standard land pattern when the maximum coincidence is obtained is specified as a part corresponding area, and the rotation angle of the standard land pattern at that time is determined. The information indicates the orientation of the part.

  Similarly, the above-described processing is executed while paying attention to all connector / land groups in order, and the processing is terminated when there are no unprocessed groups (step 913). However, regardless of which standard land pattern is used, the group type is changed to the general land group without being processed for the group for which the matching degree exceeding the threshold value cannot be obtained (steps 909 and 912). ). Therefore, the component identification process for the general land group is executed again for this group.

(E) -2 Part identification processing for IC group / general land group For the IC group and general land group, the procedure of FIG. 15 is executed for each group.

  Referring to FIG. 15, even in this process, attention is paid to one unprocessed group (step 920), and a standard land pattern corresponding to the target group is read (step 921). For example, when an IC group is focused on, a standard land pattern related to lead parts such as SOP and QFP is read, and when a general land group is focused on, a standard for a part having a pair of electrodes such as a chip part and a diode is used. The land pattern is read out.

  Next, the degree of coincidence of the land pattern of the group of interest with respect to the read standard land pattern is calculated (step 922). Here, similarly to the processing for the connector / transistor group, the position and orientation of the standard land pattern are changed, the number of pixels that are found to be coincident is counted a plurality of times, and the maximum value among them is adopted as the degree of coincidence. If the number of lands is different from the standard land pattern, the coincidence calculation is skipped.

  If the calculated coincidence exceeds the threshold value, the coincidence and the position and rotation angle of the standard land pattern when the coincidence is obtained are temporarily stored (step 924). This is also the same as the processing in the connector / transistor group.

  When the degree of coincidence is calculated using all the standard patterns corresponding to the target group, the maximum value of the degree of coincidence exceeding the threshold is specified, and the target group is based on the stored data when the maximum degree of coincidence is obtained. The component type, the component corresponding region, and the component direction corresponding to the are identified (steps 925 to 928). These processes are the same as those in the connector / transistor group.

  Similarly, the corresponding parts are specified for the IC group or the general land group, and when there is no unprocessed group (step 930), the process is terminated.

  Further, in the process for a certain group, if any of the corresponding standard land patterns is used and the degree of coincidence exceeding the threshold is not obtained, an unspecified flag is set for the group (step 929). Since it is impossible to set the next inspection data in this group, it is necessary to separately set inspection data by a user's manual operation. However, since it is considered that there is no part type data corresponding to the target group, the process proceeds to step 929. Therefore, if standard land patterns and standard inspection data are registered for a sufficient number of part types, manual operation is possible. It can be considered that the frequency of setting is negligible.

  FIG. 16 shows an example of a range including one connector / transistor group G11 and three general land groups G12, G13, G14 (corresponding to the region S in FIG. 3). An example is shown.

This region, the component identifying process (FIG. 14) for the first connector transistor group, a combination of three groups G11, G12, G13 is identified as corresponding to the standard land patterns of the transistor T _A. The remaining general land group G14 is the component identifying process for general land groups (Fig. 15), have been identified as corresponding to the standard land patterns of the chip component C _P.

  The connector / transistor group G11 is classified into the connector / transistor group by the grouping process of FIG. 10 after being classified into the read land in the land distribution process of FIG. On the other hand, the groups G12 and G13 are classified into general lands by the grouping process of FIG. 12 after being classified into general lands. In this way, even when a plurality of lands corresponding to the same part are divided into different groups, and there are land groups that do not correspond to actual parts, lands corresponding to other parts are included in each group. If not, it is finally possible to associate the groups by collating with the standard land pattern and specify the component type that matches the land pattern indicated by these groups.

Also for the three lands of transistor T _A of the central land is not the lead lands, as will be classified as a general land, if such as setting the criteria for allocation of the land, at the grouping step Three lands can be combined into one group.

(F) Inspection data setting process (step S10)
In this process, based on the result of the component specifying process, the standard inspection data is applied to the post-mounting model image to determine various inspection data including inspection area setting data. FIG. 17 shows the detailed procedure.

  Referring to FIG. 17, first, paying attention to one of the specified parts, information (part type, part corresponding area, part orientation) stored in the part specifying process is read (step 1001). Next, for the read component type, the corresponding standard inspection data is read from the standard inspection data registration unit 63 (step 1002).

  Next, with respect to the post-mounting model image and the land pattern image, the image of the component corresponding region read out in step 1001 is cut out and correlated to detect the soldered portion in the post-mounting model image (step 1003). 1004). Further, the post-mounting model image is subjected to processing such as edge extraction for the range surrounded by the detected soldering site, thereby extracting the component main body of the component of interest (step 1005).

  In this way, when the soldering part and part body of the target part are specified, the inspection area is set by applying standard inspection data based on the position and size of the part and the orientation of the part read out earlier. Then, these setting data (indicating the setting position and the size of the area) are registered in the inspection data file 61 (step 1007).

  In addition, standard inspection data for inspection data other than the inspection area, such as inspection parameters applied to each inspection area (such as the binarization threshold value and determination threshold value for detecting the inspected part), is also used. Apply and set, and register in the inspection data file 61 (step 1008).

  In the following, the inspection data of each part is set and registered by executing the same process as described above for each part specified by the part specifying process. With this, all the series of processes shown in FIG. 2 are executed, and the inspection can be executed.

FIG. 18 shows an example in which inspection areas are set for two parts specified by the part specifying process shown in FIG.
In this figure, reference numerals 301 to 306 denote inspection areas for solder inspection. These inspection areas 301 to 306 are all set for each soldering site, but the inspection areas 302 and 304 correspond to the same land (land of group G11). The inspection areas 402 and 404 are inspection areas for determining the presence / absence of a component, and are set at the center of the component main body. In addition, the inspection areas 401 and 403 are used for inspecting the suitability of the position and orientation of the component, and are set to include the entire component main body.

  FIG. 19 shows an example in which the inspection area is set based on the result after the component identification processing is executed for the IC land group set by the grouping described in FIG. In this example, the inspection areas 310 to 319 for solder inspection are set in a range including a plurality of leads and soldering portions corresponding to them. In addition, the inspection area 410 for inspecting the position and orientation of the part is set so as to include the entire part main body, as in the example of FIG. It is set near the upper right corner, not the center of the.

  The inspection areas shown in the examples of FIGS. 18 and 19 are set by applying the definitions in the standard inspection data of the corresponding component types to the component main body and the soldering part extracted from the model image after mounting. It has been done. For these inspection areas, information indicating the set position and size of the area is registered in step 1007 of FIG. 17, so that the inspection area having the same size is also placed at the same position during the inspection. It becomes possible to set.

(3) Other embodiments relating to creation of inspection data In the above embodiment, only the connector / transistor group is subjected to the matching process for the range including the adjacent group in the target group. The processing similar to that of the connector / transistor group may be performed without integrating the groups set for each array.

  In the above embodiment, land grouping is performed using three types of binary images generated from a model image of a substrate before solder application. However, the model image is not limited to the above, and the substrate after cream solder application is performed. These images may be used.

  Further, even in an image of a substrate after soldering, a high luminance area in the image can be detected as a soldering part, and a silk print pattern and a wiring pattern can also be detected. Therefore, instead of using the land of the pre-mounting model image, the high-luminance area of the post-mounting model image is detected, and these can be processed by the same method as in the above-described embodiment to identify the component. However, in this case, since only the part not hidden by the land parts is the target of grouping, there is a possibility that the setting of the group type and the algorithm for the part specifying process are slightly changed.

It is a block diagram of a board | substrate external appearance inspection apparatus. It is a schematic flowchart regarding the automatic creation process of test | inspection data. It is a figure which shows the example of the model image before mounting. It is a figure which shows the example of a land pattern image. It is a figure which shows the example of a silk pattern image. It is a figure which shows the grouping result of a land. It is a flowchart regarding the grouping of lands based on the silk inclusion area. It is a figure which shows the specific example of the process of FIG. It is a flowchart regarding land distribution processing. It is a flowchart regarding grouping of lead lands. It is a figure which shows the specific example of the integration process of IC group. It is a flowchart regarding grouping of general lands. It is a figure which shows the specific example of the process of FIG. It is a flowchart regarding the component specific process for a connector transistor group. It is a flowchart regarding the component specific process for IC groups and general land groups. It is a figure which shows the example of a components specific process. It is a flowchart regarding the setting / registration processing of inspection data. It is a figure which shows the example of a setting of a test | inspection area | region. It is a figure which shows the example of a setting of a test | inspection area | region.

Explanation of symbols

1 Camera 5 Control Unit 6 Memory 7 Image Input Unit 61 Inspection Data File 62 Land Pattern Registration Unit 63 Standard Inspection Data Registration Unit

Claims (6)

A method for creating area setting data used for setting an inspection area for each component in an image in an automatic visual inspection for a component mounting board by a computer,
A first step of registering a standard arrangement pattern of land areas corresponding to a component electrode and standard area setting data in a memory of the computer for a plurality of component types;
A second step of individually detecting land areas, silk print patterns and wiring patterns on the board from the image of the model board before or after component mounting, and generating three types of binary images representing the respective detection results; ,
A third step of grouping the land areas based on the correspondence relationship of the three types of binary images, provided that a set of land areas having a relationship involving silk print patterns or wiring patterns is not classified into the same group; ,
After the grouping is completed, the process of matching the binary image of the land area with the standard arrangement pattern of the land area for each part type is executed on condition that the land area in the same group is associated with the same part type. A fourth step of identifying a component type, a mounting range, and a component orientation of each component mounted on the substrate;
For each component specified in the fourth step, an inspection to be set for the component by applying standard area setting data of the component type specified based on the orientation of the component specified in the specified mounting range. A fifth step of identifying an area and registering area setting data for setting the inspection area in the memory of the computer;
A method for creating region setting data for an inspection region, characterized in that:   The method for creating region setting data for an inspection region according to claim 1, wherein, in the third step, one group is set by land regions arranged in a state surrounded by a silk print pattern.   In the third step, a plurality of land regions whose areas and shapes are aligned are arranged in the same group and arranged along a certain direction without any silk printing pattern or wiring pattern interposed therebetween. The method for creating region setting data for an inspection region according to claim 1.   2. The method according to claim 1, wherein, in the third step, a plurality of land regions whose areas and shapes are matched and arranged without any silk printing pattern and wiring pattern are classified into the same group. How to create area setting data for the inspection area.   2. The inspection according to claim 1, wherein in the fourth step, a standard pattern of one component type is associated with a distribution pattern of land areas of one or a plurality of groups to obtain the number of land areas and the degree of arrangement matching. How to create area setting data for an area. Imaging means for imaging a substrate, and area setting data creating means for creating area setting data of an inspection area to be set for each component in an inspection target image generated by imaging the substrate to be inspected by the imaging means And a memory for storing the created area setting data, and setting an inspection area on the image to be inspected based on the area setting data registered in the memory, and processing an image in each inspection area on the substrate In a board appearance inspection apparatus comprising: an inspection execution unit that determines a mounting state of each component; and an output unit that outputs a determination result by the inspection execution unit.
In the memory, for a plurality of component types, standard inspection data including standard layout patterns of land regions corresponding to component electrodes and standard region setting data of inspection regions are registered,
The area setting data creation means includes
Binary image generation that individually detects land areas, silk-printed patterns, and wiring patterns on the board from the model board image before or after component mounting, and generates three types of binary images representing the detection results Means,
Grouping means for grouping each land area based on the correspondence relationship of the three types of binary images, provided that the set of land areas in a relationship involving silk print patterns or wiring patterns is not classified into the same group;
After the grouping is completed, the process of matching the binary image of the land area with the standard arrangement pattern of the land area for each part type is executed on condition that the land area in the same group is associated with the same part type. Component specifying means for specifying the component type, mounting range, and component orientation of each component mounted on the board;
For each component specified by the component specifying means, the standard area setting data of the component type specified based on the direction of the component specified in the specified mounting range is applied to the component. A substrate visual inspection apparatus comprising: a registration unit that specifies an inspection area and registers area setting data for setting the inspection area in the memory.

Images