High-density circuit board and method based on rapid heat dissipation of chip

Abstract

The present application discloses a high-density circuit board and method for rapid heat dissipation of a chip. The high-density circuit board has a chip installation area and a chip heat dissipation element installation area. The chip installation area and the chip heat dissipation element installation area are adapted to install chips and heat dissipation elements, and the chip heat dissipation element installation area is located near the chip installation area. In particular, the method uses machine vision technology to achieve distortion correction and rotation correction of the drilling image, so as to avoid the problem that the spatial domain offset between the image and the reference image caused by the shooting angle makes the two images unable to be compared and analyzed by correlation analysis. This helps to more intelligently achieve quality correction and control of the high-density circuit board to ensure the processing quality of the circuit board.

Description

Technical Field

[0001] The present application relates to the field of intelligent correction, and more specifically, to a high-density circuit board and method based on rapid heat dissipation of chips. Background Art

[0002] With the continuous progress of technology, electronic devices are rapidly developing towards smaller size, higher performance, and higher integration. This trend poses higher requirements for the design and manufacturing of circuit boards, especially the growing demand for high-density circuit boards. In particular, high-density circuit boards can provide more circuit connections, adapt to more complex electronic system designs, and meet the dual requirements of devices for space utilization and performance.

[0003] Currently, due to the relatively high circuit layout density of high-density circuit boards, their heat dissipation function is worse than that of traditional circuit boards. Some existing technologies set up a chip heat dissipation element installation area near the chip installation area of the high-density circuit board to achieve rapid heat dissipation of the high-density circuit board chips by installing chip heat dissipation elements in this area. However, due to the introduction of chip heat dissipation elements, more attention needs to be paid to the processing details of high-density circuit boards, which poses higher requirements for the drilling process of high-density circuit boards to avoid damage to the circuit board caused by incorrect or deviated opening positions of the high-density circuit board. Chinese Patent Publication CN116847554A proposes a quality control system, method, device, and medium for high-density circuit boards, which locates the circuit board through a bearing platform, and the processor controls a laser emitter to perform precise material removal based on drilling data. The sensing device collects drilling images and analyzes the drilling conditions of the circuit board based on the gray information of the drilling images to generate correction data. Finally, the processor generates control instructions according to the correction data to command the laser emitter to perform drilling correction to ensure that the drilling quality of the circuit board meets high-standard requirements.

[0004] However, in the above-mentioned quality control system for high-density circuit boards, it determines the correction data by performing gray analysis on the drilling images and looking up the error correction table. Only using image gray analysis may not fully utilize the implicit information and semantic association relationships in the images, especially when dealing with complex or noisy drilling images. In addition, the correction method based on looking up the table may be effective for specific types of errors, but it is not very adaptable to other types of errors or new unknown errors. Moreover, the method of looking up the table usually requires comparing and analyzing each drilling image one by one, which will reduce the processing speed and is rather cumbersome.

[0005] Therefore, an optimized method for high-density circuit boards based on rapid heat dissipation of chips is desired, which can achieve drilling quality correction and control of high-density circuit boards in a more intelligent way to avoid damage to the circuit board caused by incorrect or deviated opening positions of the circuit board. Summary of the Invention

[0006] To solve the above technical problems, the present application is proposed. Embodiments of the present application provide a high-density circuit board and method based on rapid heat dissipation of chips, which capture the features of drilling images and enhance the significant features through image recognition and analysis technology based on machine vision, so as to generate a drilling image after rotation correction and achieve distortion and rotation correction of the drilling image. Furthermore, the corrected drilling image is subjected to pixel-by-pixel difference with the drilling reference image to automatically generate the correction data. In this way, it is possible to use machine vision technology to achieve distortion correction and rotation correction of drilling images, avoiding the problem that the spatial domain offset between the image and the reference image caused by the shooting angle makes the two images unable to be compared and analyzed for correlation. This helps to more intelligently achieve quality correction and control of high-density circuit boards to ensure the processing quality of the circuit boards.

[0007] According to one aspect of the present application, there is provided a method for a high-density circuit board based on rapid heat dissipation of chips, which includes:

[0008] Place the circuit board at a specific position on the support platform and collect drilling data information;

[0009] Based on the drilling data information, the central processing unit controls the laser emission device to emit a first laser beam to remove the material layer in the drilling area of the circuit board;

[0010] Collect the drilling image through the sensing component;

[0011] Perform feature analysis on the drilling image based on the significant enhancement mechanism of the image foreground mask to generate correction data;

[0012] Based on the correction data, the central processing unit controls the laser emission device to emit a second laser beam to perform drilling correction on the drilling area.

[0013] Compared with the prior art, the high-density circuit board and method based on rapid heat dissipation of chips provided by the present application capture the features of drilling images and enhance the significant features through image recognition and analysis technology based on machine vision, so as to generate a drilling image after rotation correction and achieve distortion and rotation correction of the drilling image. Furthermore, the corrected drilling image is subjected to pixel-by-pixel difference with the drilling reference image to automatically generate the correction data. In this way, it is possible to use machine vision technology to achieve distortion correction and rotation correction of drilling images, avoiding the problem that the spatial domain offset between the image and the reference image caused by the shooting angle makes the two images unable to be compared and analyzed for correlation. This helps to more intelligently achieve quality correction and control of high-density circuit boards to ensure the processing quality of the circuit boards. Brief Description of the Drawings

[0014] The above and other objects, features, and advantages of the present application will become more apparent by describing the embodiments of the present application in more detail with reference to the accompanying drawings. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. They are used together with the embodiments of the present application to explain the present application and do not constitute a limitation to the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0015] Figure 1 FIG. is a flowchart of a method for a high-density circuit board based on rapid heat dissipation of a chip according to an embodiment of the present application;

[0016] Figure 2 FIG. is a schematic diagram of data flow of a method for a high-density circuit board based on rapid heat dissipation of a chip according to an embodiment of the present application;

[0017] Figure 3 FIG. is a flowchart of sub-step S4 of a method for a high-density circuit board based on rapid heat dissipation of a chip according to an embodiment of the present application. Detailed Embodiments

[0018] Hereinafter, exemplary embodiments of the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments of the present application. It should be understood that the present application is not limited by the exemplary embodiments described herein.

[0019] As shown in the present application and the claims, unless the context clearly indicates an exception, words such as "a", "an", "one", and / or "the" are not specifically singular and may also include plural. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of the clearly identified steps and elements, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0020] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules may be used and run on the user terminal and / or the server. The modules are merely illustrative, and different aspects of the system and method may use different modules.

[0021] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the operations before or below do not necessarily need to be executed precisely in order. On the contrary, as needed, various steps may be executed in reverse order or simultaneously. At the same time, other operations may also be added to these processes, or one or several steps may be removed from these processes.

[0022] Next, exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all embodiments of the present application. It should be understood that the present application is not limited by the exemplary embodiments described herein.

[0023] The drilling quality on a circuit board is crucial for ensuring the precise assembly of electronic components and the stable operation of the circuit. The accuracy, alignment, and integrity of the hole walls of the drill holes are directly related to the electrical performance and mechanical strength of the circuit board. Currently, due to the relatively high circuit layout density of high-density circuit boards, their heat dissipation function is worse than that of traditional circuit boards. Some existing technologies set a chip heat dissipation element installation area near the chip installation area of the high-density circuit board to achieve rapid heat dissipation of the chips on the high-density circuit board by installing chip heat dissipation elements in this area. However, due to the introduction of chip heat dissipation elements, more attention needs to be paid to the processing details of high-density circuit boards, which poses higher requirements for the drilling process of high-density circuit boards to avoid damage to the circuit board caused by incorrect or deviated opening positions of the high-density circuit board. The drilling correction operation is an indispensable part of circuit board manufacturing, which is directly related to the performance, reliability, and production cost of electronic products. Therefore, an optimized method for high-density circuit boards based on rapid chip heat dissipation is desired, which can achieve the correction and control of the drilling quality of high-density circuit boards in a more intelligent way to avoid damage to the circuit board caused by incorrect or deviated opening positions of the circuit board.

[0024] In the technical solution of the present application, a method for high-density circuit boards based on rapid chip heat dissipation is proposed. Figure 1 FIG. is a flowchart of a method for high-density circuit boards based on rapid chip heat dissipation according to an embodiment of the present application. Figure 2 FIG. is a schematic diagram of data flow of a method for high-density circuit boards based on rapid chip heat dissipation according to an embodiment of the present application. As Figure 1 and Figure 2 shown, the method for high-density circuit boards based on rapid chip heat dissipation according to an embodiment of the present application includes the steps of: S1, placing the circuit board at a specific position on the support platform and collecting drilling data information; S2, based on the drilling data information, controlling a laser emission device to emit a first laser beam through a central processing unit to remove the material layer in the drilling area of the circuit board; S3, collecting a drilling image through a sensing component; S4, performing feature analysis on the drilling image based on an image foreground mask significant enhancement mechanism to generate correction data; S5, based on the correction data, the central processing unit controls the laser emission device to emit a second laser beam to perform drilling correction on the drilling area of the circuit board.

[0025] Specifically, in step S1, the circuit board is placed at a specific position on the support platform, and drilling data information is collected. It should be understood that in this step, the circuit board is not the final formed high-density circuit board but the circuit board in the process. For the convenience of distinction, in some embodiments, the circuit board can be defined as the initial circuit board, but this is not limited to the present application. On the other hand, the drilling data information includes drilling depth, diameter, position, etc. By comparing the actual drilling data with the reference drilling data, any deviation can be identified and corrected, thereby improving the overall quality of the circuit board.

[0026] Specifically, in step S2, based on the drilling data information, the central processing unit controls the laser emission device to emit a first laser beam to remove the material layer in the drilling area of the circuit board. It should be understood that the removal of the material layer reduces the thermal resistance between the circuit board and the surrounding environment. By reducing the thermal resistance, the heat dissipation efficiency can be improved.

[0027] Specifically, in step S3, a drilling image is collected by the sensing component. Among them, the drilling image may include information such as drilling position, drilling depth, drilling diameter, etc. The drilling image can be used for operations such as visualizing drilling data and drilling correction.

[0028] Specifically, in step S4, feature analysis based on the image foreground mask significant enhancement mechanism is performed on the drilling image to generate correction data. Specifically, in a specific example of the present application, as Figure 3 shown, step S4 includes: S41, extracting a drilling reference image from the database; S42, performing distortion correction on the drilling image to obtain a corrected drilling image; S43, inputting the corrected drilling image into a drilling morphology feature extractor to obtain a drilling morphology feature map; S44, inputting the drilling morphology feature map into a feature foreground mask significanter based on the convolutional gated feedforward mechanism to obtain a foreground significant drilling morphology feature map; S45, performing rotation correction on the foreground significant drilling morphology feature map to obtain a rotation-corrected drilling image; S46, calculating the per-pixel position difference between the rotation-corrected drilling image and the drilling reference image to obtain a drilling difference pixel matrix; S47, based on the drilling difference pixel matrix, obtaining the correction data.

[0029] Specifically, in step S41, a drilling reference image is extracted from the database. It should be understood that the drilling reference image contains information about the area to be drilled and drilling parameters. The drilling reference image is used to guide the drilling process to ensure that the drilling is accurate and meets the specifications.

[0030] Specifically, in step S42, the drilling image is subjected to distortion correction to obtain a corrected drilling image. Considering that the drilling image is acquired by a sensing component, it may be distorted due to lens distortion, shooting angle, or the irregularity of the circuit board itself. Based on this, in the technical solution of this application, the drilling image is subjected to distortion correction to obtain a corrected drilling image, which can improve the accuracy of the drilling image, make the details in the image clearer, and thus improve the accuracy of subsequent image analysis and processing.

[0031] Specifically, in step S43, the corrected drilling image is input into a drilling morphology feature extractor to obtain a drilling morphology feature map. In a specific example of this application, the corrected drilling image is input into a drilling morphology feature extractor based on a dilated convolutional neural network model to obtain the drilling morphology feature map. Considering that the dilated convolutional neural network has good perception and capture capabilities for image feature extraction and analysis. And the corrected drilling image contains key feature information of the drilling morphology. Based on this, in the technical solution of this application, the corrected drilling image is input into a drilling morphology feature extractor based on a dilated convolutional neural network model to capture and extract the detailed features of the drilling morphology and obtain a drilling morphology feature map. It is worth mentioning that the dilated convolutional neural network (DCNN) is a type of convolutional neural network (CNN) that uses dilated convolution operations to expand the receptive field while maintaining the spatial resolution. Here, the receptive field refers to the area covered by the convolutional kernel in the input image.

[0032] Specifically, in step S44, the drilling morphology feature map is input into a feature foreground mask saliency detector based on a convolutional gated feedforward mechanism to obtain a foreground salient drilling morphology feature map. Considering that different drilling morphology features in the drilling morphology feature map express different degrees of feature information, which reflects the most critical and salient feature information of the drilling morphology, but there is also irrelevant background noise. Therefore, in order to selectively focus on and strengthen the importance information of specific regions in the drilling morphology feature map while reducing the influence of features with lower contribution degrees. In the technical solution of this application, the drilling morphology feature map is input into a feature foreground mask saliency detector based on a convolutional gated feedforward mechanism to obtain a foreground salient drilling morphology feature map. It should be understood that the feature foreground mask saliency detector based on the convolutional gated feedforward mechanism highlights the key drilling morphology feature information through adaptive dynamic feature selection based on a gated mask for the drilling morphology feature map.

[0033] In an embodiment of the present application, inputting the drilling morphology feature map into a feature foreground mask saliency detector based on a convolutional gated feedforward mechanism to obtain a foreground salient drilling morphology feature map includes: first performing layer normalization processing on the drilling morphology feature map to obtain a normalized drilling morphology feature map; that is, introducing layer normalization to perform layer normalization processing on the drilling morphology feature map to reduce internal covariate shift and make the input distribution of each layer more stable. Next, performing channel expansion based on point convolution and depth convolution encoding based on a dilated convolutional layer on the normalized drilling morphology feature map to obtain a drilling morphology depth convolution original feature map and a drilling morphology depth convolution backup feature map; by performing point convolution and dilated convolution encoding on the normalized features, the expression ability of the features can be increased, and features of more complex structures can be captured to obtain the drilling morphology depth convolution original feature map and the drilling morphology depth convolution backup feature map respectively. Then, inputting the drilling morphology depth convolution original feature map into a foreground gated mask module based on function to obtain a drilling morphology depth convolution gated mask weight feature map; here, inputting the convolution original feature map into the Gelu function for activation processing to better express and distinguish different features, improve the recognition ability of the features, and then performing foreground salient mask processing on the activated feature map so that the model can adaptively select and focus on the most important and salient parts of the drilling features, filtering out irrelevant or background information to obtain the drilling morphology depth convolution gated mask weight feature map. Further, calculating the point-by-position multiplication between the drilling morphology depth convolution gated mask weight feature map and the drilling morphology depth convolution backup feature map to obtain a drilling morphology gated mask foreground highlighting feature map; by calculating the point-by-position multiplication between the drilling morphology depth convolution gated mask weight feature map and the drilling morphology depth convolution backup feature map, the salient foreground features in the feature map are highlighted. Finally, performing channel contraction based on point convolution on the drilling morphology gated mask foreground highlighting feature map to obtain the foreground salient drilling morphology feature map. That is, performing point convolution encoding on the highlighted features to complete the final feature channel compression to obtain the foreground salient drilling morphology feature map.

[0034] Among them, inputting the drilling morphology depth convolution original feature map into a foreground gated mask module based on function to obtain a drilling morphology depth convolution gated mask weight feature map includes: inputting the drilling morphology depth convolution original feature map into function activation function to obtain a drilling morphology depth convolution original activation feature map; performing masking processing based on a foreground gated mechanism on the drilling morphology depth convolution original activation feature map to obtain the drilling morphology depth convolution gated mask weight feature map.

[0035] More specifically, a masking process based on a foreground gating mechanism is performed on the original activation feature map of the borehole morphology depth convolution to obtain the weighted feature map of the borehole morphology depth convolution gating mask, including: setting the feature values greater than a predetermined hyperparameter at each position of the original activation feature map of the borehole morphology depth convolution to one, and the rest to zero to obtain the weighted feature map of the borehole morphology depth convolution gating mask.

[0036] In summary, in the above embodiment, the borehole morphology feature map is input into the feature foreground mask saliency detector based on the convolutional gating feedforward mechanism to obtain the foreground salient borehole morphology feature map, including: inputting the borehole morphology feature map into the feature foreground mask saliency detector based on the convolutional gating feedforward mechanism, and processing it with the following foreground mask saliency calculation formula to obtain the foreground salient borehole morphology feature map; wherein, the foreground mask saliency calculation formula is:

[0037]

[0038]

[0039]

[0040]

[0041]

[0042]

[0043]

[0044] Wherein, is the borehole morphology feature map, represents the layer normalization operation on the feature map, is the normalized borehole morphology feature map, is the point convolution operation, is the dilated convolution operation with a convolution kernel of , is the original feature map of the borehole morphology depth convolution, is the backup feature map of the borehole morphology depth convolution, is the activation function, is the masking process, is the weighted feature map of the borehole morphology depth convolution gating mask, is the feature value at each position in the feature map, is the predetermined hyperparameter, is the pointwise multiplication by position, is the foreground salient feature map of the borehole morphology gating mask, It is the foreground significant drilling morphology feature map.

[0045] Specifically, in S45, the foreground significant drilling morphology feature map is rotationally corrected to obtain a rotationally corrected drilling image. In a specific example of this application, the foreground significant drilling morphology feature map is input into a rotation correction generator based on a diffusion model to obtain the rotationally corrected drilling image. Considering that there may be a spatial domain offset between the drilling image and the reference image due to different shooting angles during the acquisition of the drilling image, therefore, in order to generate a high-quality and high-resolution image based on the foreground significant drilling morphology feature map, so as to more clearly reflect the detailed features of the drilling, in the technical solution of this application, the foreground significant drilling morphology feature map is input into a rotation correction generator based on a diffusion model to obtain a rotationally corrected drilling image. That is to say, the diffusion model is good at processing complex image generation and conversion tasks, and generates high-quality images through a step-by-step denoising process. Specifically, by rotationally correcting the foreground significant drilling morphology feature map, subtle drilling morphology features and structural information can be effectively captured, thereby generating a more accurate rotationally corrected drilling image. In this way, the image deformation (such as barrel distortion or pillow distortion) caused by the optical characteristics of the sensor and lens distortion can be corrected through machine vision technology and rotation correction technology, and at the same time, the spatial domain offset problem between the drilling image and the reference image caused by the shooting angle can be solved to ensure image alignment, which helps to more accurately identify and correct the deviation of the drilling, and realize the quality correction and control of high-density circuit boards.

[0046] Specifically, in S46, the pixel-by-pixel position difference between the rotationally corrected drilling image and the drilling reference image is calculated to obtain a drilling difference pixel matrix. That is, in the technical solution of this application, the pixel-by-pixel position difference between the rotationally corrected drilling image and the drilling reference image is calculated to obtain a drilling difference pixel matrix, so as to reveal the possible small errors or deviations during the correction process, including differences in position, shape, and size, etc., providing rich data decision support for the generation of subsequent correction data.

[0047] Specifically, in S47, the correction data is obtained based on the drilled differential pixel matrix. In a specific example of the present application, the drilled differential pixel matrix is input into a correction data generator based on a decoder to obtain the correction data. That is, the drilled differential pixel matrix obtained by performing pixel value differential calculation between the rotation-corrected drilled image and the drilled reference image is decoded to automatically generate the correction data. In this way, it is possible to correct the image deformation (such as barrel distortion or pillow distortion) caused by the optical characteristics of the sensor and lens distortion through machine vision technology and rotation correction technology, and at the same time solve the spatial domain offset problem between the image and the reference image caused by the shooting angle, ensuring image alignment. And the significant features of the drilled image are extracted using image recognition and analysis technology, which helps to more accurately identify and correct the deviation of the drill hole, improving the processing accuracy.

[0048] In a preferred example, the applicant of the present application considers that the foreground significant drilled morphology feature map represents the convolutional gated foreground saliency image semantic features of the drilled image. Therefore, after the foreground significant drilled morphology feature map is input into a rotation correction generator based on a diffusion model, the obtained rotation-corrected drilled image will also have an uneven diffusion generation corresponding to the spatial distribution difference of the image semantic features, resulting in insufficient coverage of the image semantic difference representation in the drilled differential pixel matrix obtained by calculating the per-pixel position difference between it and the drilled reference image, causing an outlier regression inference mapping deviation when inputting into a correction data generator based on a decoder for decoding regression, affecting the accuracy of the correction data obtained by inputting the drilled differential pixel matrix into a correction data generator based on a decoder.

[0049] Based on this, when the drilled differential pixel matrix is input into a correction data generator based on a decoder to obtain the correction data, the drilled differential pixel matrix is optimized. The optimization process specifically includes:

[0050] Calculate the characteristic mean of the drilled differential pixel matrix, and divide the characteristic mean by the difference between the maximum eigenvalue and the minimum eigenvalue of the drilled differential pixel matrix to obtain a drilled differential pixel distribution characterization value;

[0051] Divide one minus the drilled differential pixel distribution characterization value by the drilled differential pixel distribution characterization value to obtain a drilled differential pixel distribution modulation value;

[0052] Activate the drilled differential pixel matrix through a probability function to obtain a probabilistic drilled differential pixel matrix;

[0053] After subtracting the probabilistic drilled differential pixel matrix from the drilled differential pixel distribution modulation value, take the absolute value and calculate the negative logarithm with base 2 to obtain a probabilistic drilled differential pixel distribution modulation information matrix;

[0054] Divide the value representing the distribution of drilled hole differential pixels by one minus the difference of each eigenvalue of the probabilized drilled hole differential pixel matrix, sum over all eigenvalues of the probabilized drilled hole differential pixel matrix, and divide by the scale of the drilled hole differential pixel matrix to obtain the modulation bias value of the probabilized drilled hole differential pixel distribution; and

[0055] Dot multiply the probabilized drilled hole differential pixel distribution modulation information matrix and the modulation bias value of the probabilized drilled hole differential pixel distribution by the weight as a hyperparameter to obtain the optimized drilled hole differential pixel matrix.

[0056] Among them, the optimization of the drilled hole differential pixel matrix is expressed as:

[0057]

[0058]

[0059] Among them, is the probabilized drilled hole differential pixel matrix, is the value representing the distribution of drilled hole differential pixels, is pointwise subtraction by position, represents the logarithmic function value with base 2, is pointwise addition by position, is the weight as a hyperparameter, is the scale of the drilled hole differential pixel matrix, i.e., width times height, is each eigenvalue of the probabilized drilled hole differential pixel matrix, is the mean eigenvalue of the drilled hole differential pixel matrix, and are the maximum eigenvalue and the minimum eigenvalue of the drilled hole differential pixel matrix respectively, is the optimized drilled hole differential pixel matrix.

[0060] That is, in the above preferred example, the probability information distribution planning of the drilled differential pixel matrix based on pixel values is performed through the Bernoulli probability modulation distribution of the drilled differential pixel matrix with respect to the pixel value distribution, and the probability inverse mapping of the overall probability characteristics of the drilled differential pixel matrix is used as the extended coverage of the set mapping space of the drilled differential pixel matrix, so as to independently understand the intuitive probability information distribution and the abstract probability space mapping of the drilled differential pixel matrix, and to improve the accuracy of the corrected data obtained by inputting the drilled differential pixel matrix into the decoder-based corrected data generator by avoiding the outlier feature distribution of the drilled differential pixel matrix to the counterfactual inference mapping of the decoded regression probability. In this way, the image deformation (such as barrel distortion or pillow distortion) caused by the optical characteristics of the sensor and lens distortion can be corrected through machine vision technology and rotation correction technology, and at the same time, the spatial domain offset problem between the image and the reference image caused by the shooting angle can be solved to ensure image alignment. And the significant features of the drilled hole image are extracted by using image recognition and analysis technology, which helps to more accurately identify and correct the deviation of the drilled hole and improve the processing accuracy.

[0061] It is worth mentioning that in other specific examples of the present application, the drilled hole image can also be analyzed for features based on the image foreground mask significant enhancement mechanism in other ways to generate corrected data. For example: input the drilled hole image; use an image segmentation algorithm (such as U-Net or Mask R-CNN) to generate the foreground mask of the drilled hole image. Among them, the foreground mask is a binary image, where the foreground area (drilled hole) is 1 and the background area is 0; apply the foreground mask to the drilled hole image to only enhance the foreground area; use image enhancement techniques (such as histogram equalization, sharpening or edge detection) to enhance the foreground area; combine the enhanced foreground area with the original background area to generate corrected data.

[0062] Specifically, in step S5, based on the corrected data, the central processing unit controls the laser emission device to emit a second laser beam to perform drilling correction on the drilling area of the circuit board. In this way, the quality correction and control of the high-density circuit board are realized to ensure the processing quality of the circuit board.

[0063] In summary, the method of a high-density circuit board based on rapid chip heat dissipation according to the embodiments of the present application is elucidated. By adopting image recognition and analysis techniques based on machine vision, feature capture and significant feature enhancement processing of the drilling image are carried out to generate a drilling image after rotation correction, thereby realizing distortion and rotation correction of the drilling image. Furthermore, the corrected drilling image is subjected to pixel-by-pixel difference with the drilling reference image to automatically generate the correction data. In this way, the machine vision technology can be used to realize distortion correction and rotation correction of the drilling image, so as to avoid the problem that the spatial domain offset between the image and the reference image caused by the shooting angle makes the two images unable to be compared and analyzed by correlation. This helps to more intelligently realize the quality correction and control of the high-density circuit board to ensure the processing quality of the circuit board.

[0064] The embodiments of the present disclosure have been described above. The above description is exemplary and not exhaustive, and is not limited to the disclosed embodiments. Many modifications and variations are obvious to those of ordinary skill in the art in the technical field without departing from the scope and spirit of the described embodiments. The selection of the terms used herein is intended to best explain the principles of the embodiments, practical applications, or improvements to the technology in the market, or to enable other ordinary skilled persons in the technical field to understand the embodiments disclosed herein.

Claims

1. A method for a high-density circuit board based on rapid heat dissipation of a chip, wherein the high-density circuit board has a chip mounting area and a chip heat dissipation element mounting area, wherein the chip mounting area and the chip heat dissipation element mounting area are suitable for mounting a chip and a heat dissipation element, and the chip heat dissipation element mounting area is located near the chip mounting area, characterized in that: The method comprises: Place the circuit board at a specific position on the support platform and collect drilling data information; Based on the drilling data information, a central processing unit is used to control a laser emitting device to emit a first laser beam to remove a material layer in a drilling area of ​​the circuit board; Acquire drilling images through a sensor component; Performing feature analysis on the borehole image based on an image foreground mask significant enhancement mechanism to generate correction data; Based on the correction data, the central processing unit controls the laser emitting device to emit a second laser beam to perform drilling correction on the drilling area of ​​the circuit board; The step of performing feature analysis on the drilling image based on an image foreground mask significant enhancement mechanism to generate correction data includes: extracting a drilling reference image from a database; Performing distortion correction on the borehole image to obtain a corrected borehole image; Inputting the corrected borehole image into a borehole morphology feature extractor to obtain a borehole morphology feature map; Inputting the drilling morphology feature map into a feature foreground mask salient device based on a convolutional gated feed-forward mechanism to obtain a foreground salient drilling morphology feature map; Performing rotation correction on the foreground significant borehole morphological feature map to obtain a rotation-corrected borehole image; Calculating the pixel-by-pixel position difference between the rotation-corrected drilling image and the drilling reference image to obtain a drilling differential pixel matrix; The correction data is obtained based on the drilling differential pixel matrix.

2. The method for high-density circuit board based on rapid heat dissipation of chips according to claim 1, characterized in that: Inputting the corrected borehole image into a borehole morphological feature extractor to obtain a borehole morphological feature map, including: inputting the corrected borehole image into a borehole morphological feature extractor based on a hole convolutional neural network model to obtain the borehole morphological feature map.

3. The method for high-density circuit board based on rapid heat dissipation of chips according to claim 2, characterized in that: Inputting the drilling morphology feature map into a feature foreground mask salient device based on a convolution gated feedforward mechanism to obtain a foreground salient drilling morphology feature map, including: Performing layer normalization processing on the borehole morphology feature map to obtain a normalized borehole morphology feature map; Performing point convolution-based channel expansion and hole convolution-layer-based deep convolution coding on the normalized drilling morphology feature map to obtain a drilling morphology deep convolution original feature map and a drilling morphology deep convolution backup feature map; The drilling morphology deep convolution original feature map is input based on The foreground gated mask module of the function is used to obtain the drilling morphology deep convolution gated mask weight feature map; Calculate the position point multiplication between the drilling morphology deep convolution gated mask weight feature map and the drilling morphology deep convolution backup feature map to obtain a drilling morphology gated mask foreground highlighting feature map; The drilling morphology gated mask foreground salient feature map is subjected to point convolution-based channel shrinkage to obtain the foreground significant drilling morphology feature map.

4. The method for high-density circuit board based on rapid heat dissipation of chips according to claim 3, characterized in that: The drilling morphology deep convolution original feature map is input based on The foreground gated mask module of the function is used to obtain the drilling morphology deep convolution gated mask weight feature map, including: Input the drilling morphology deep convolution original feature map Function activation function to obtain the original activation feature map of the drilling morphology deep convolution; The drilling morphology deep convolution original activation feature map is masked based on the foreground gating mechanism to obtain the drilling morphology deep convolution gated mask weight feature map.

5. The method for high-density circuit board based on rapid heat dissipation of chips according to claim 4, characterized in that: The drilling morphology deep convolution original activation feature map is subjected to masking processing based on a foreground gating mechanism to obtain the drilling morphology deep convolution gated mask weight feature map, including: The feature values ​​greater than the predetermined hyperparameter in each position of the drilling morphology deep convolution original activation feature map are set to one, and the rest are set to zero to obtain the drilling morphology deep convolution gated mask weight feature map.

6. The method for high-density circuit board based on rapid heat dissipation of chips according to claim 5, characterized in that: The foreground significant borehole morphological feature map is rotationally corrected to obtain a rotation-corrected borehole image, comprising: inputting the foreground significant borehole morphological feature map into a rotation-corrected generator based on a diffusion model to obtain the rotation-corrected borehole image.

7. The method for high-density circuit board based on rapid heat dissipation of chips according to claim 6, characterized in that: Based on the drilled differential pixel matrix, the correction data is obtained, comprising: inputting the drilled differential pixel matrix into a correction data generator based on a decoder to obtain the correction data.

Images