Integrated circuit board printing method and device

Abstract

The invention discloses an integrated circuit board printing method and device, and the method comprises the steps: obtaining an amplified image set of the surface of a plurality of substrates from an image collection sequence, the image set comprising high-resolution images of a plurality of regions of the substrates, and the images at least comprising a first image; calculating and analyzing a plurality of first images based on the weight map, and extracting first image values from the first images through a feature extraction algorithm; according to the substrate area where the first image value is located and the three-dimensional model in the substrate area, at least one coordinate is constructed, a coordinate system is generated according to the multiple coordinates, and at least one three-dimensional appearance first texture is continuously generated through the coordinate system; in response to the first texture of the three-dimensional appearance, the intelligent analysis model is adopted to obtain the first texture, and the first texture is intelligently matched with a database texture model; and if the matching fails, sending a prompt message through the intelligent analysis model. The three-dimensional morphology of the surface of the integrated circuit board can be detected, defects of the three-dimensional morphology can be found in time to complete processing, and the production quality of the integrated circuit board is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of circuit board manufacturing, and in particular to an integrated circuit board printing method and device. Background Art

[0002] Integrated circuit board printing usually refers to the manufacturing process of printed circuit boards. This process involves converting circuit designs into actual physical circuit boards so that various electronic components such as integrated circuits, resistors, capacitors, etc. can be installed and connected on them. With the rapid development of the electronics manufacturing industry and the rapid improvement of industrial automation levels, the existing printed circuit board manufacturing is rapidly developing towards high performance, high density and high complexity, which not only increases the manufacturing difficulty, but also makes the existing PCB defect detection methods face many problems and challenges.

[0003] Usually, before an integrated circuit board is generated, it needs to go through the preliminary circuit schematic and PCB layout design, and then go through processes such as making photolithography films, pre-processing substrates, lamination, etching, electroplating, silk screen printing, etc. Finally, it needs to pass appearance and electrical performance tests to ensure that there are no defects such as short circuits or open circuits. Once the inspection passes, a complete integrated circuit board can be generated. However, due to the current technology, there are deviations in the three-dimensional morphology of the surface during the generation of integrated circuit boards. If the defects are not discovered in time, it will be difficult to accurately detect solder joint defects and component mounting postures, resulting in poor production quality of integrated circuit boards. Therefore, it is urgent to set up a method and device that is convenient for quickly and accurately detecting the three-dimensional morphology of the surface of integrated circuit boards to solve the above problems. Summary of the invention

[0004] The purpose of the present invention is to provide an integrated circuit board printing method and device, which can quickly and accurately detect the three-dimensional morphology of the surface of the integrated circuit board during the integrated circuit board generation process, timely discover the defects of the three-dimensional morphology and complete the processing, thereby improving the production quality of the integrated circuit board.

[0005] The technical solution of the present invention is achieved in this way:

[0006] A first aspect of the present invention provides an integrated circuit board printing method, comprising:

[0007] Acquire a plurality of magnified image collections of the substrate surface from an image acquisition sequence, wherein the magnified image collection includes high-resolution images of multiple regions of the substrate, and the images include at least one first image;

[0008] Calculating and analyzing the plurality of first images based on the weight map, and extracting the first image value from the first images by using a feature extraction algorithm;

[0009] constructing at least one coordinate according to the substrate region where the first image value is located and a three-dimensional model in the substrate region, wherein a coordinate system is generated according to a plurality of the coordinates, and at least one first texture with a three-dimensional appearance is continuously generated with the coordinate system;

[0010] In response to the first texture of the three-dimensional appearance, an intelligent analysis model is used to obtain the first texture and perform intelligent matching with a database texture model; if the matching fails, a prompt message is issued through the intelligent analysis model.

[0011] A further technical solution is that before acquiring a collection of a plurality of enlarged images of the substrate surface from the image acquisition sequence, generating the image acquisition sequence includes:

[0012] Using a light generator to project light onto the surface of the substrate;

[0013] Acquire multiple exposure images by combining a high-speed optical camera module with the projection light generated by the light generator;

[0014] The exposure image is enlarged by an image amplifier to form a plurality of high-resolution image atlases.

[0015] A further technical solution is that the calculation and analysis of the plurality of first images based on the weight map, and extracting the first image value from the first image by a feature extraction algorithm, includes:

[0016] Using graph theory to set edges for the target object in the enlarged image;

[0017] The edge line is assigned a first weight by a Pearson coefficient;

[0018] The first weight is used to identify the edge of the target object in the enlarged image using a Sobel operator to generate the first image value.

[0019] A further technical solution is that the first image value includes the first weight of each edge line of the three-dimensional model in the substrate area.

[0020] A further technical solution is that constructing at least one coordinate according to the substrate region where the first image value is located and the three-dimensional model in the substrate region includes:

[0021] jointly optimizing the parameters of the first image value using a standard gauge block;

[0022] In the three-dimensional model, any first image value adopts Zhang Zhengyou calibration method to construct at least one three-dimensional coordinate.

[0023] A further technical solution is that a coordinate system is generated according to the plurality of coordinates, and at least one three-dimensional appearance first texture is continuously generated with the coordinate system, wherein the three-dimensional appearance first texture includes:

[0024] The coordinate system includes a plurality of first three-dimensional coordinates, and the plurality of first three-dimensional coordinates are connected to form the three-dimensional appearance first texture;

[0025] The three-dimensional appearance first texture includes a straight line appearance texture and a curve appearance texture.

[0026] A further technical solution is that, in response to the first texture of the three-dimensional appearance, an intelligent analysis model is used to obtain the first texture and perform intelligent matching with a database texture model, including:

[0027] The intelligent analysis model obtains the straight line appearance texture and the curve appearance texture data;

[0028] Matching the straight line appearance texture and the curve appearance texture with the preset values ​​in the database through the intelligent analysis model;

[0029] If the match is successful, the detection is completed;

[0030] If the matching fails, a prompt message is issued through the intelligent analysis model.

[0031] A further technical solution is that the cluster analysis is used to match the straight line texture and the curve texture with the preset value in the database, including:

[0032] grouping each standardized texture in the database into the same cluster;

[0033] The straight line appearance texture or the curved line appearance texture is generated into a straight line appearance texture cluster and a curved line appearance texture cluster by using the cluster analysis model;

[0034] Inputting the straight line appearance texture cluster and the curved line appearance texture cluster into the database and confirming the same cluster similar to the straight line appearance texture cluster and the curved line appearance texture cluster;

[0035] The straight line appearance texture cluster and the curve appearance texture cluster are completed to be the same cluster as the one in the database, and it is determined whether the straight line appearance texture cluster and the curve appearance texture cluster are the same.

[0036] A second aspect of the present invention provides an integrated circuit board printing device, comprising:

[0037] An image storage unit, used for acquiring a collection of multiple magnified images of the substrate surface from an image acquisition sequence, wherein the collection of magnified images includes high-resolution images of multiple regions of the substrate, and the images include at least a first image value;

[0038] An assignment unit, configured to calculate and analyze a collection of the plurality of enlarged images based on a weight map, and extract the first image value from the collection by a feature extraction algorithm;

[0039] A texture generating unit is used to construct at least one coordinate according to the substrate region where the first image value is located and a three-dimensional model in the substrate region, wherein a coordinate system is generated according to a plurality of the coordinates, and at least one three-dimensional appearance first texture is continuously generated with the coordinate system;

[0040] A matching unit is used to respond to the first texture of the three-dimensional appearance, use an intelligent analysis model to obtain the first texture and perform intelligent matching with a database texture model; if the matching fails, a prompt message is issued through the intelligent analysis model.

[0041] A further technical solution is to also include: an image acquisition unit, which is used to project light onto the surface of the substrate using a light generator; obtain multiple exposure images through a high-speed optical camera module combined with the projection light generated by the light generator; and amplify the exposure images through an image amplifier to form multiple high-resolution image sets.

[0042] The beneficial effects of the present invention are:

[0043] The image acquisition sequence stores the surface magnified images of the substrate to be inspected, and the magnified images are all high-resolution images, which are convenient for subsequent analysis. One of the multiple images at the same position in the substrate area is selected as the first image, and the first image is analyzed and calculated through the weight map, and then the first image is assigned a first image value through the feature extraction method. According to the substrate area where the first image value is located and the three-dimensional model in the substrate area, it is convenient to confirm the specific position for mounting components and construct coordinates. The first image value can be assigned with multiple coordinates, and then a coordinate system is generated, which is convenient for the computer to describe the three-dimensional appearance first texture according to the coordinate system. After the three-dimensional appearance first texture is obtained through the intelligent analysis model, it is compared with the preset value in the database. If the match fails, a prompt message is issued through the intelligent analysis model to prompt the staff to reconstruct the three-dimensional appearance of the substrate, thereby solving the problem that the surface three-dimensional morphology of the integrated circuit board is deviated during the generation process. If the defects are not discovered in time, it is not conducive to accurately detecting solder joint defects and component mounting postures, resulting in poor production quality of integrated circuit boards. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 is a flow chart of the method of the present invention;

[0045] Figure 2 It is a block diagram of the device system of the present invention. DETAILED DESCRIPTION

[0046] In order to better understand the technical content of the present invention, specific embodiments are provided below, and the present invention is further described in conjunction with the accompanying drawings.

[0047] In view of the background technology, in the current technology, there are deviations in the three-dimensional surface morphology during the generation process of integrated circuit boards. If the defects are not discovered in time, it will be difficult to accurately detect solder joint defects and component mounting postures, resulting in poor production quality of integrated circuit boards.

[0048] To solve the above problems, an embodiment of the present invention provides an integrated circuit board printing method, in which an enlarged image of the surface of the substrate to be inspected is stored in the image acquisition sequence, and the enlarged images are all high-resolution images, which are convenient for subsequent analysis. One of the multiple images of the same position in the substrate area is selected as the first image, and the first image is analyzed and calculated through the weight map, and then the first image is assigned a first image value through the feature extraction method. According to the substrate area where the first image value is located and the three-dimensional model in the substrate area, it is convenient to confirm the specific position for mounting components and construct coordinates. The first image value can be assigned with multiple coordinates, and then a coordinate system is generated, which is convenient for the computer to describe the three-dimensional appearance first texture according to the coordinate system. After the three-dimensional appearance first texture is obtained by the intelligent analysis model, it is compared with the preset value in the database. If the match fails, a prompt message is issued through the intelligent analysis model to prompt the staff to reconstruct the three-dimensional appearance of the substrate, thereby solving the problem that the surface three-dimensional morphology is deviated during the generation process of the integrated circuit board. If the defects are not discovered in time, it is not conducive to accurately detecting solder joint defects and component mounting postures, resulting in poor production quality of the integrated circuit board.

[0049] See also Figure 1 The first aspect of the embodiment of the present invention provides an integrated circuit board printing method, which mainly includes the following steps:

[0050] S101: Acquire a collection of multiple enlarged images of a substrate surface from an image acquisition sequence, wherein the collection of enlarged images includes high-resolution images of multiple regions of the substrate, and the images include at least one first image.

[0051] In S101, generating the image acquisition sequence includes: using a light generator to project onto the substrate surface; acquiring multiple exposure images through a high-speed optical camera module combined with the projection light generated by the light generator; and amplifying the exposure images through an image amplifier to form multiple high-resolution image atlases.

[0052] The light generator can be used to project pure white light, and multiple groups of light generators at different angles can be set at the same time. When the light is projected, the subtle parts of the three-dimensional model on the substrate can be illuminated, so that the high-speed optical camera module has no shadows during the shooting process, thereby improving the image quality. The high-speed optical camera module is a camera component specially designed to capture high-speed motion scenes. They can capture images at very high frame rates so that these images can be clearly observed. The image amplifier can enhance the details of the three-dimensional appearance of the substrate surface by analyzing the texture and structure of the image through super-resolution reconstruction technology.

[0053] S102: Calculate and analyze the plurality of first images based on the weight map, and extract the first image values ​​from the first images using a feature extraction algorithm.

[0054] In S102, graph theory is used to set edges for the target object in the enlarged image; the edges are assigned a first weight through the Pearson coefficient; the first weight is used to identify the edges of the target object in the enlarged image using the Sobel operator to generate the first image value.

[0055] In the principle of graph theory, the target object in the enlarged image is a three-dimensional appearance, wherein the edge of the three-dimensional appearance is set as a node through the corner, and a line is set between the two nodes as the edge. First, for each group of target objects (i.e., the edge presented by the three-dimensional appearance in the enlarged image), the Pearson correlation coefficient is calculated. The Pearson coefficient r ranges from -1 to +1, where +1 indicates a complete positive correlation, -1 indicates a complete negative correlation, and 0 indicates no linear relationship. Secondly, a threshold t is set to determine which Pearson coefficients are large enough to form an edge. Finally, the Pearson coefficient with an absolute value greater than t is calculated to establish the edge. For example, if the absolute value of the Pearson coefficient between two nodes in the three-dimensional appearance in the enlarged image is greater than t, a edge is created between them, and this coefficient is used as the weight of the edge, and the first weight is assigned. The Sobel operator is used to detect the edge, and the gradient of the enlarged image is calculated by convolution operation, so as to find the edge of the corner in the image, which is used to assign the first image value. Wherein, the first image value includes the first weight of each edge of the three-dimensional model in the substrate area. The Sobel operator detects edges so that computers can clearly distinguish the edges of three-dimensional appearances.

[0056] S103: Construct at least one coordinate according to the substrate region where the first image value is located and a three-dimensional model in the substrate region, wherein a coordinate system is generated according to a plurality of the coordinates, and at least one three-dimensional first texture is continuously generated with the coordinate system.

[0057] In S103, the parameters of the first image value are jointly optimized using a standard measuring block; any first image value in the three-dimensional model is calibrated using the Zhang Zhengyou calibration method to construct at least one three-dimensional coordinate; the coordinate system includes multiple first three-dimensional coordinates, and the multiple first three-dimensional coordinates are connected to form the three-dimensional appearance first texture; the three-dimensional appearance first texture includes a straight line appearance texture and a curve appearance texture.

[0058] The standard gauge block is a reference standard or baseline. The numerical value in the first image value is accurate to three decimal places to complete the joint optimization, so that the computer can clearly and accurately describe the coordinates of the three-dimensional appearance in the enlarged image. The edge lines in the same direction are the first three-dimensional coordinates, and the edges in different directions are the second three-dimensional coordinates and the third three-dimensional coordinates respectively. Multiple first three-dimensional coordinates or second three-dimensional coordinates or third three-dimensional coordinates in the same direction can continuously form a first texture or a second texture or a third texture. In the first texture, multiple first three-dimensional coordinates include but are not limited to different X-axes, different Y-axes or one of the Z-axes, or two of the three quadrants are different, or all three quadrants are different. Specifically, it is expressed as a straight texture, a straight zigzag texture or a curved texture. Similarly, the second three-dimensional coordinates and the third three-dimensional coordinates also have this situation.

[0059] S104: In response to the first texture of the three-dimensional appearance, an intelligent analysis model is used to obtain the first texture and perform intelligent matching with a database texture model; if the matching fails, a prompt message is issued through the intelligent analysis model.

[0060] In S104, the intelligent analysis model obtains the straight line appearance texture and the curve appearance texture data; the straight line appearance texture and the curve appearance texture are matched with the preset values ​​in the database through the intelligent analysis model; if the match is successful, the detection is completed; if the match fails, a prompt message is issued through the intelligent analysis model.

[0061] The intelligent analysis model receives an image containing straight line appearance texture and curved line appearance texture data, and compares it with the standardized data in the database including the specific values ​​of the three-dimensional coordinates, and compares the straight line appearance texture and curved line appearance texture data one by one. If the straight line appearance texture and curved line appearance texture data match the specific values ​​of the three-dimensional coordinates in the database, it is a normal three-dimensional appearance. Specifically, the straight line texture includes different Y axes but the same X and Z axes (for example, the first group: X = 0.125, Y = 0.122, Z = 0.110, the second group: X = 0.125, Y = 0.120, Z = 0.110), if there is the same matching data in the database, it is a normal value; if the line texture includes different X and Y axes but the same Z axis (for example, the first group: X = 0.125, Y = 0.129, Z = 0.110, the second group: X = 0.125, Y = 0.120, Z = 0.110), there is no such straight line data in the database, it is an abnormal value, and the intelligent analysis model sends a prompt message to prompt the staff to reconstruct the three-dimensional appearance of the substrate. However, if it is a straight curve, it is a normal value. It is worth noting that the data must be completely the same to be a normal value. In addition, when comparing the curve appearance texture, since the three-dimensional coordinates of the first group are different from the three-dimensional coordinates of the second group, when comparing, the matching must be completed in accordance with the principle of complete data identity. If they are different, they are abnormal values.

[0062] It should be noted that the prompt message is expressed in text on the display screen, specifically the difference in three-dimensional coordinates, so that the staff can intuitively see the difference. The subsequent reconstruction process can make normal reconstruction based on the difference, thereby improving work efficiency.

[0063] Furthermore, the clustering analysis is used to match the straight line appearance texture and the curved line appearance texture with the preset values ​​in the database, including: grouping each standardized texture into the same cluster in the database; generating a straight line appearance texture cluster and a curved line appearance texture cluster from the straight line appearance texture or the curved line appearance texture through the clustering analysis model; inputting the straight line appearance texture cluster and the curved line appearance texture cluster into the database to confirm the same cluster that is similar to the straight line appearance texture cluster and the curved line appearance texture cluster; completing the straight line appearance texture cluster and the curved line appearance texture cluster with the same cluster in the database, and determining whether the straight line appearance texture cluster and the curved line appearance texture cluster are the same.

[0064] Cluster analysis can be used. Cluster analysis can help us automatically group data such as the first group of the first three-dimensional coordinates, as well as the second group, the third group and other groups into the same cluster, and then match them with preset values ​​in the database to achieve rapid identification.

[0065] See also Figure 2 According to a second aspect of an embodiment of the present invention, there is provided an integrated circuit board printing device, comprising:

[0066] An image storage unit is used to obtain a collection of multiple enlarged images of the substrate surface from an image acquisition sequence, wherein the collection of enlarged images includes high-resolution images of multiple regions of the substrate, and the images include at least one first image value; an assignment unit is used to calculate and analyze the collection of multiple enlarged images based on a weight map, and extract the first image value from the collection through a feature extraction algorithm; a texture generation unit is used to construct at least one coordinate according to the substrate region where the first image value is located and a three-dimensional model in the substrate region, wherein a coordinate system is generated according to the multiple coordinates, and at least one three-dimensional appearance first texture is continuously generated with the coordinate system; a matching unit is used to respond to the three-dimensional appearance first texture, use an intelligent analysis model to obtain the first texture and perform intelligent matching with a database texture model; if the matching fails, a prompt message is issued through the intelligent analysis model. An image acquisition unit is used to project on the substrate surface using a light generator; obtain multiple exposure images through a high-speed optical camera module combined with the projection light generated by the light generator; and amplify the exposure images through an image amplifier to form multiple high-resolution image atlases.

[0067] It should be noted that the working principle and effect of the present device are consistent with those of the above method, and will not be described in detail here. Among them, the present device and method are both based on an electric control system to complete the automatic control process.

[0068] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for printing an integrated circuit board, characterized in that: include: Acquire a plurality of magnified image collections of the substrate surface from an image acquisition sequence, wherein the magnified image collection includes high-resolution images of multiple regions of the substrate, and the images include at least one first image; Calculating and analyzing the plurality of first images based on the weight map, and extracting the first image value from the first images by using a feature extraction algorithm; constructing at least one coordinate according to the substrate region where the first image value is located and a three-dimensional model in the substrate region, wherein a coordinate system is generated according to a plurality of the coordinates, and at least one first texture with a three-dimensional appearance is continuously generated with the coordinate system; In response to the first texture of the three-dimensional appearance, an intelligent analysis model is used to obtain the first texture and perform intelligent matching with a database texture model; if the matching fails, a prompt message is issued through the intelligent analysis model.

2. The integrated circuit board printing method according to claim 1, characterized in that: Before acquiring a collection of a plurality of enlarged images of the substrate surface from the image acquisition sequence, generating the image acquisition sequence includes: Using a light generator to project light onto the surface of the substrate; Acquire multiple exposure images by combining a high-speed optical camera module with the projection light generated by the light generator; The exposure image is enlarged by an image amplifier to form a plurality of high-resolution image atlases.

3. The integrated circuit board printing method according to claim 1, characterized in that: The step of calculating and analyzing the plurality of first images based on the weight map, and extracting the first image value from the first images by using a feature extraction algorithm, includes: Using graph theory to set edges for the target object in the enlarged image; The edge line is assigned a first weight by a Pearson coefficient; The first weight is used to identify the edge of the target object in the enlarged image using a Sobel operator to generate the first image value.

4. The integrated circuit board printing method according to claim 3, characterized in that: The first image value includes the first weight of each of the edge lines of the three-dimensional model in the substrate region.

5. The integrated circuit board printing method and device according to claim 1, characterized in that: The step of constructing at least one coordinate according to the substrate region where the first image value is located and the three-dimensional model in the substrate region comprises: jointly optimizing the parameters of the first image value using a standard gauge block; In the three-dimensional model, any first image value adopts Zhang Zhengyou calibration method to construct at least one three-dimensional coordinate.

6. The integrated circuit board printing method according to claim 1, characterized in that: The step of generating a coordinate system according to the plurality of coordinates, and continuously generating at least one three-dimensional appearance first texture with the coordinate system, wherein the three-dimensional appearance first texture comprises: The coordinate system includes a plurality of first three-dimensional coordinates, and the plurality of first three-dimensional coordinates are connected to form the three-dimensional appearance first texture; The three-dimensional appearance first texture includes a straight line appearance texture and a curve appearance texture.

7. The integrated circuit board printing method according to claim 6, characterized in that: In response to the first texture of the three-dimensional appearance, using an intelligent analysis model to obtain the first texture and perform intelligent matching with a database texture model, includes: The intelligent analysis model obtains the straight line appearance texture and the curve appearance texture data; Matching the straight line appearance texture and the curve appearance texture with the preset values ​​in the database through the intelligent analysis model; If the match is successful, the detection is completed; If the matching fails, a prompt message is issued through the intelligent analysis model.

8. The integrated circuit board printing method according to claim 7, characterized in that: The step of matching the straight line texture and the curve texture with preset values ​​in the database by cluster analysis includes: Grouping each standardized texture into the same cluster in the database; The straight line appearance texture or the curved line appearance texture is generated into a straight line appearance texture cluster and a curved line appearance texture cluster by using the cluster analysis model; Inputting the straight line appearance texture cluster and the curved line appearance texture cluster into the database and confirming the same cluster similar to the straight line appearance texture cluster and the curved line appearance texture cluster; The straight line appearance texture cluster and the curve appearance texture cluster are completed with the same cluster in the database, and it is determined whether the straight line appearance texture cluster and the curve appearance texture cluster are the same.

9. An integrated circuit board printing device, using the method according to any one of claims 1 to 8, characterized in that: include: An image storage unit, used for acquiring a collection of multiple magnified images of the substrate surface from an image acquisition sequence, wherein the collection of magnified images includes high-resolution images of multiple regions of the substrate, and the images include at least a first image value; An assignment unit, configured to calculate and analyze a collection of the plurality of enlarged images based on a weight map, and extract the first image value from the collection by a feature extraction algorithm; A texture generating unit is used to construct at least one coordinate according to the substrate region where the first image value is located and a three-dimensional model in the substrate region, wherein a coordinate system is generated according to a plurality of the coordinates, and at least one three-dimensional appearance first texture is continuously generated with the coordinate system; A matching unit is used to respond to the first texture of the three-dimensional appearance, use an intelligent analysis model to obtain the first texture and perform intelligent matching with a database texture model; if the matching fails, a prompt message is issued through the intelligent analysis model.

10. The integrated circuit board printing device according to claim 9, characterized in that: Also includes: An image acquisition unit, used for projecting light onto the surface of the substrate using a light generator; A plurality of exposure images are acquired by combining a high-speed optical camera module with the projection light generated by the light generator; the exposure images are amplified by an image amplifier to form a plurality of high-resolution image atlases.

Images