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

By using machine vision technology to capture and significantly enhance the feature of drilling images on high-density circuit boards, the problems of insufficient heat dissipation and poor quality control in the existing technology are solved, and intelligent quality correction for high-density circuit boards are achieved and control, improving processing accuracy and speed.

CN118973101AActive Publication Date: 2024-11-15RED BOARD JIANGXI CO LTD
View PDF 9 Cites 0 Cited by

Patent Information

Application Number
CN202411158535.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-22
Publication Date
2024-11-15
Estimated Expiration
2044-08-22

AI Technical Summary

Technical Problem

The existing high-density circuit boards are relatively lacking in heat dissipation function, and the existing quality control system has insufficient adaptability and slow processing speed through image grayscale analysis and table lookup correction methods, making it difficult to effectively avoid errors in the circuit board opening position or Damage caused by deviation.

Method used

Using machine vision-based image recognition and analysis technology to capture and remarkable feature enhancement processing of drilling images, generate rotation-corrected drilling images, and generate correction data by pixel-by-pixel difference to realize distortion and rotation of drilling images Calibrate to ensure the quality of the circuit board processing.

Benefits of technology

Through machine vision technology, intelligent quality correction and control of high-density circuit boards is achieved, processing accuracy and speed are improved, image offset problems caused by shooting angles are avoided, and high-quality production of circuit boards is ensured.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118973101A_ABST
    Figure CN118973101A_ABST
Patent Text Reader

Abstract

The invention discloses a high-density circuit board and method based on rapid heat dissipation of a chip, the high-density circuit board is provided with a chip installation area and a chip heat dissipation element installation area, and the chip installation area and the chip heat dissipation element installation area are suitable for installing chips and heat dissipation elements. And the chip heat dissipation element mounting area is positioned near the chip mounting area. Particularly, according to the method, distortion correction and rotation correction of a drilling image are achieved through the machine vision technology, so that the problem that the two images cannot be compared and subjected to correlation analysis due to spatial domain offset between the image and a reference image caused by a shooting angle is solved; therefore, the quality correction and control of the high-density circuit board can be realized more intelligently, so that the processing quality of the circuit board is ensured.
Need to check novelty before this filing date? Find Prior Art

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 advancement of technology, electronic devices are rapidly developing towards smaller, higher performance and higher integration. This trend has put forward higher requirements for circuit board design and manufacturing, 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 needs of equipment for space utilization and performance.

[0003] At present, due to the relatively high circuit layout density of high-density circuit boards, their heat dissipation function is poorer 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 high-density circuit board chip by installing the chip heat dissipation element in this area. However, due to the introduction of chip heat dissipation elements, more attention needs to be paid to the processing details of the high-density circuit board, which puts higher requirements on the drilling process of the high-density circuit board to avoid damage to the circuit board due to errors or deviations in the opening position of the high-density circuit board. China's public patent CN116847554A proposes a quality control system, method, device and medium for high-density circuit boards, which positions the circuit board through a carrying platform, and the processor controls the laser transmitter to perform precise material removal based on the drilling data. The sensing device collects the drilling image and analyzes the drilling condition of the circuit board based on the grayscale information of the drilling image, thereby generating correction data. Finally, the processor generates a control instruction based on the correction data, directs the laser transmitter to perform drilling correction, and ensures that the drilling quality of the circuit board meets high standards.

[0004] However, in the quality control system of the high-density circuit board mentioned above, the correction data is determined by grayscale analysis of the drilling image and error correction lookup table. Using only image grayscale analysis may not fully utilize the implicit information and semantic associations in the image, especially when dealing with complex or noisy drilling images. In addition, the correction method based on table lookup may be effective for specific types of errors, but is not adaptable to other types of errors or new unknown errors. Moreover, the table lookup method usually requires each drilling image to be compared and analyzed one by one, which will reduce the processing speed and is more cumbersome.

[0005] Therefore, an optimized method for high-density circuit boards based on rapid chip heat dissipation is desired, which can achieve correction and control of the drilling quality of high-density circuit boards in a more intelligent way to avoid damage to the circuit boards due to errors or deviations in the opening position of the circuit boards. Summary of the invention

[0006] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides a high-density circuit board and method based on rapid heat dissipation of chips, which uses image recognition and analysis technology based on machine vision to perform feature capture and significant feature enhancement processing of drilling images, so as to generate a drilling image after rotation correction, and realize distortion and rotation correction of the drilling image. Then, the corrected drilling image is differentiated from the drilling reference image pixel by pixel to automatically generate the correction data. In this way, 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 it impossible to compare and correlate the two images, which helps to realize the quality correction and control of high-density circuit boards more intelligently, so as to ensure the processing quality of the circuit boards.

[0007] According to one aspect of the present application, a method for high-density circuit board based on rapid heat dissipation of chips is provided, comprising: Place the circuit board at a specific position on the support platform and collect drilling data information; Based on the drilling data information, the central processing unit controls the laser emitting device to emit a first laser beam to remove the material layer in the 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.

[0008] According to another aspect of the present application, a high-density circuit board based on rapid chip heat dissipation is provided, and the high-density circuit board device based on rapid chip heat dissipation is used to execute the high-density circuit board method based on rapid chip heat dissipation according to claims 1-8.

[0009] Compared with the prior art, the present application provides a high-density circuit board and method based on rapid heat dissipation of chips, which uses image recognition and analysis technology based on machine vision to perform feature capture and significant feature enhancement processing of drilling images, so as to generate a drilling image after rotation correction, and realize distortion and rotation correction of the drilling image. Then, the corrected drilling image is differentiated from the drilling reference image pixel by pixel to automatically generate the correction data. In this way, 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 it impossible to compare and correlate the two images, which helps to realize the quality correction and control of high-density circuit boards more intelligently, so as to ensure the processing quality of the circuit boards. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0011] Figure 1 It is a flow chart of a method for a high-density circuit board based on rapid heat dissipation of chips according to an embodiment of the present application; Figure 2 A data flow diagram of a method for a high-density circuit board based on rapid heat dissipation of chips according to an embodiment of the present application; Figure 3 This is a flowchart of sub-step S4 of the method for a high-density circuit board based on rapid chip heat dissipation according to an embodiment of the present application. DETAILED DESCRIPTION

[0012] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0013] As shown in this application and claims, unless the context clearly indicates an exception, the words "a", "an", "an" and / or "the" do not refer to the singular and may also include the plural. Generally speaking, the terms "include" and "comprise" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0014] 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 can be used and run on the user terminal and / or server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0015] 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 preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps may be processed in reverse order or simultaneously as required. Meanwhile, other operations may also be added to these processes, or a certain step or several steps of operations may be removed from these processes.

[0016] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0017] The quality of drilling holes on circuit boards is the key to ensuring accurate assembly of electronic components and stable operation of circuits. The accuracy, alignment and integrity of the drilling holes are directly related to the electrical performance and mechanical strength of the circuit board. At present, due to the relatively high circuit layout density of high-density circuit boards, their heat dissipation function is poorer than that of traditional circuit boards. Some existing technologies are to set a chip heat dissipation component installation area near the chip mounting area of ​​the high-density circuit board to achieve rapid heat dissipation of high-density circuit board chips by installing chip heat dissipation components in this area. However, due to the introduction of chip heat dissipation components, more attention needs to be paid to the processing details of high-density circuit boards, which puts higher requirements on the drilling process of high-density circuit boards to avoid damage to the circuit board due to errors or deviations in the opening position of the high-density circuit board. The drilling correction operation is an indispensable link in the manufacture of circuit boards, which is directly related to the performance, reliability and production cost of electronic products. Therefore, it is expected that an optimized method for high-density circuit boards based on rapid heat dissipation of chips can realize 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 due to errors or deviations in the opening position of the circuit board.

[0018] In the technical solution of the present application, a method for a high-density circuit board based on rapid heat dissipation of chips is proposed. Figure 1 The present invention is a flowchart of a method for a high-density circuit board based on rapid heat dissipation of chips according to an embodiment of the present application. Figure 2 The data flow diagram of the method for high-density circuit board based on rapid heat dissipation of chips according to an embodiment of the present application is shown in FIG. Figure 1 and Figure 2As shown, according to the embodiment of the present application, the method for a high-density circuit board based on rapid heat dissipation of chips includes the following steps: S1, placing the circuit board at a specific position of a support platform and collecting drilling data information; S2, based on the drilling data information, controlling the laser emitting device through a central processing unit to emit a first laser beam 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, controlling the central processing unit to control the laser emitting device to emit a second laser beam to perform drilling correction on the drilling area of ​​the circuit board.

[0019] In particular, in S1, the circuit board is placed at a specific position of the support platform and drilling data information is collected. It should be understood that in this step, the circuit board is not a final high-density circuit board but a circuit board in the process. For the sake of distinction, in some embodiments, the circuit board can be defined as a preliminary circuit board, which is not limited by this application. On the other hand, the drilling data information includes drilling depth, diameter, and 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.

[0020] In particular, in S2, based on the drilling data information, the central processing unit controls the laser emitting device to emit a first laser beam to remove the material layer of 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, and by reducing the thermal resistance, the heat dissipation efficiency can be improved.

[0021] In particular, the S3 collects drilling images through the sensor component, wherein the drilling images may include information such as drilling position, drilling depth, and drilling diameter, and the drilling images can be used for operations such as visualizing drilling data and drilling correction.

[0022] In particular, the step S4 performs feature analysis on the drilling image based on the image foreground mask significant enhancement mechanism to generate correction data. In particular, in a specific example of the present application, Figure 3As shown, the S4 includes: S41, extracting a drilling reference image from a database; S42, performing distortion correction on the drilling image to obtain a corrected drilling image; S43, inputting the corrected drilling image into a drilling morphological feature extractor to obtain a drilling morphological feature map; S44, inputting the drilling morphological feature map into a feature foreground mask salient device based on a convolutional gated feedforward mechanism to obtain a foreground-significant drilling morphological feature map; S45, performing rotation correction on the foreground-significant drilling morphological feature map to obtain a rotation-corrected drilling image; S46, 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; S47, obtaining the correction data based on the drilling differential pixel matrix.

[0023] Specifically, the S41 extracts a drilling reference image from a 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.

[0024] Specifically, in S42, the drilling image is subjected to distortion correction to obtain a corrected drilling image. Considering that the drilling image is acquired through a sensor component, it may be distorted due to lens distortion, shooting angle or irregularity of the circuit board itself. Based on this, in order to ensure that the features in the drilling image can be correctly identified and measured, in the technical solution of the present 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.

[0025] Specifically, in S43, the corrected borehole image is input into a borehole morphology feature extractor to obtain a borehole morphology feature map. In a specific example of the present application, the corrected borehole image is input into a borehole morphology feature extractor based on a hole convolutional neural network model to obtain the borehole morphology feature map. Considering that the hole convolutional neural network has good perception and capture capabilities for feature extraction and analysis of images. The corrected borehole image contains key feature information of the borehole morphology. Based on this, in the technical solution of the present application, the corrected borehole image is input into a borehole morphology feature extractor based on a hole convolutional neural network model to capture and mine out the detailed features of the borehole morphology, and obtain a borehole morphology feature map. It is worth mentioning that a hole convolutional neural network (DCNN) is a convolutional neural network (CNN) that uses a hole convolution operation to expand the receptive field while maintaining spatial resolution. Among them, the receptive field refers to the area covered by the convolution kernel in the input image.

[0026] Specifically, in S44, the borehole morphology feature map is input into a feature foreground mask salient device based on a convolution gated feedforward mechanism to obtain a foreground-significant borehole morphology feature map. Considering that different borehole morphology features in the borehole morphology feature map express different degrees of feature information, which reflects the most critical and significant feature information of the borehole morphology, but there is also irrelevant background noise. Therefore, in order to selectively focus on and strengthen the importance information of a specific area in the borehole morphology feature map, while reducing the influence of features with lower contribution. In the technical solution of the present application, the borehole morphology feature map is input into a feature foreground mask salient device based on a convolution gated feedforward mechanism to obtain a foreground-significant borehole morphology feature map. It should be understood that the feature foreground mask salient device based on a convolution gated feedforward mechanism highlights the key borehole morphology feature information by performing adaptive dynamic feature selection based on a gated mask on the borehole morphology feature map.

[0027] In an embodiment of the present application, the drilling morphology feature map is input into a feature foreground mask salient device based on a convolutional gated feedforward mechanism to obtain a foreground-salient drilling morphology feature map, including: firstly 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, so that the input distribution of each layer is more stable. Next, the normalized drilling morphology feature map is subjected to channel expansion based on point convolution and deep convolution encoding based on a hole convolution layer to obtain a drilling morphology deep convolution original feature map and a drilling morphology deep convolution backup feature map; by performing point convolution and hole convolution encoding on the normalized features, the expressive power of the features can be increased, and features of more complex structures can be captured to obtain the drilling morphology deep convolution original feature map and the drilling morphology deep convolution backup feature map, respectively. Then, the drilling morphology deep convolution original feature map is input into a feature foreground mask salient device based on a convolutional gated feedforward mechanism to obtain a foreground-salient drilling morphology deep convolution original feature map and a drilling morphology deep convolution backup feature map. The foreground gated mask module of the function is used to obtain the drilling morphology deep convolution gated mask weight feature map; here, the convolution original feature map is input into the Gelu function for activation processing to better express and distinguish different features and improve the feature recognition ability, and then the feature map obtained after activation is processed with foreground saliency mask so that the model can adaptively select and focus on the most important and significant parts of the drilling features, and filter out irrelevant or background information to obtain the drilling morphology deep convolution gated mask weight feature map. Further, the position-based point multiplication between the drilling morphology deep convolution gated mask weight feature map and the drilling morphology deep convolution backup feature map is calculated to obtain the drilling morphology gated mask foreground highlight feature map; by calculating the position-based point multiplication between the drilling morphology deep convolution gated mask weight feature map and the drilling morphology deep convolution backup feature map, the significant foreground features in the feature map are highlighted. Finally, the drilling morphology gated mask foreground highlight feature map is subjected to channel contraction based on point convolution to obtain the foreground significant drilling morphology feature map. That is, the highlighted features are point-convolutionally encoded to complete the final feature channel compression, and the foreground-significant drilling morphology feature map is obtained.

[0028] The drilling morphology deep convolution original feature map is input based on The foreground gated mask module of the function is used to obtain a drilling morphology deep convolution gated mask weight feature map, including: inputting the drilling morphology deep convolution original feature map into A function activation function is used to obtain a drilling morphology deep convolution original activation feature map; and 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.

[0029] More specifically, the drill morphology deep convolution original activation feature map is masked based on the foreground gating mechanism to obtain the drill morphology deep convolution gated mask weight feature map, including: setting the feature values ​​greater than a predetermined hyperparameter in each position of the drill morphology deep convolution original activation feature map to one, and setting the rest to zero to obtain the drill morphology deep convolution gated mask weight feature map.

[0030] In summary, in the above embodiment, the drilling morphology feature map is input into a feature foreground mask salient device based on a convolution gated feedforward mechanism to obtain a foreground significant drilling morphology feature map, including: inputting the drilling morphology feature map into a feature foreground mask salient device based on a convolution gated feedforward mechanism, and processing the foreground significant drilling morphology feature map using the following foreground mask saliency calculation formula; wherein the foreground mask saliency calculation formula is:

[0031]

[0032]

[0033]

[0034]

[0035]

[0036] in, is the characteristic diagram of the drilling morphology, Indicates the layer normalization operation on the feature map. is the normalized drilling morphology characteristic map, is the point convolution operation, The convolution kernel is The dilated convolution operation, is the original feature map of the drilling morphology deep convolution, The deep convolution backup feature map of the drilling morphology is for Activation function, For masking, is the drilling morphology deep convolution gated mask weight feature map, is the feature value of each position in the feature map, To predetermine the hyperparameters, For point-by-position multiplication, The foreground highlight feature map of the drilling morphology gated mask, A diagram showing the significant borehole morphological features of the prospect.

[0037] Specifically, the S45 performs rotation correction on the foreground significant borehole morphological feature map to obtain a rotation-corrected borehole image. In a specific example of the present application, the foreground significant borehole morphological feature map is input into a rotation correction generator based on a diffusion model to obtain the rotation-corrected borehole image. Considering that there may be a spatial domain offset between the borehole image and the reference image due to different shooting angles when the borehole image is collected, in order to generate a high-quality, high-resolution image based on the foreground significant borehole morphological feature map, so as to more clearly reflect the detailed features of the borehole, in the technical solution of the present application, the foreground significant borehole morphological feature map is input into a rotation correction generator based on a diffusion model to obtain a rotation-corrected borehole image. That is, 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 performing rotation correction on the foreground significant borehole morphological feature map, subtle borehole morphological features and structural information can be effectively captured, thereby generating a more accurate rotation-corrected borehole image. In this way, machine vision technology and rotation correction technology can be used to correct image deformations (such as barrel distortion or pincushion distortion) caused by sensor optical characteristics and lens distortion, while solving the spatial domain offset problem between the drilling image and the reference image caused by the shooting angle to ensure image alignment, which helps to more accurately identify and correct drilling deviations and achieve quality correction and control of high-density circuit boards.

[0038] Specifically, the S46 calculates the pixel-by-pixel position difference between the rotation-corrected drilling image and the drilling reference image to obtain a drilling differential pixel matrix. That is, in the technical solution of the present application, the pixel-by-pixel position difference between the rotation-corrected drilling image and the drilling reference image is calculated to obtain a drilling differential pixel matrix, so as to reveal the possible minor errors or deviations in the correction process, including differences in position, shape, size, etc., and provide rich data decision support for the generation of subsequent correction data.

[0039] Specifically, the S47 obtains the correction data based on the drilling differential pixel matrix. In a specific example of the present application, the drilling differential pixel matrix is ​​input into a correction data generator based on a decoder to obtain the correction data. That is, the drilling differential pixel matrix obtained by performing pixel value difference calculation between the rotation-corrected drilling image and the drilling reference image is decoded to automatically generate the correction data. In this way, the image deformation (such as barrel distortion or pincushion distortion) caused by the optical characteristics of the sensor and the lens distortion can be corrected by machine vision technology and rotation correction technology, while solving the spatial domain offset problem between the image and the reference image caused by the shooting angle to ensure image alignment. And the use of image recognition and analysis technology to extract the significant features of the drilling image helps to more accurately identify and correct the deviation of the drilling and improve the processing accuracy.

[0040] In a preferred example, the applicant of the present application takes into account that the foreground-significant drilling morphological feature map represents the convolution-gated foreground-salient image semantic features of the drilling image. Therefore, after the foreground-significant drilling morphological feature map is input into a rotation-corrected generator based on a diffusion model, the obtained rotation-corrected drilling image will also have a diffusion generation imbalance corresponding to the difference in the spatial distribution of image semantic features, resulting in insufficient image semantic differential representation coverage in the drilling differential pixel matrix obtained by calculating the pixel-by-pixel position difference between it and the drilling reference image, causing an outlier regression inference mapping deviation when the decoding regression is input into a decoder-based correction data generator, affecting the accuracy of the corrected data obtained by inputting the drilling differential pixel matrix into the decoder-based correction data generator.

[0041] Based on this, when the drilled differential pixel matrix is ​​input into the decoder-based correction data generator to obtain the correction data, the drilled differential pixel matrix is ​​optimized, and the optimization process specifically includes: Calculating a characteristic mean of the drilling differential pixel matrix, and dividing the characteristic mean by a difference between a maximum eigenvalue and a minimum eigenvalue of the drilling differential pixel matrix to obtain a drilling differential pixel distribution characterization value; Subtracting the drilling differential pixel distribution characterization value from one and dividing it by the drilling differential pixel distribution characterization value to obtain a drilling differential pixel distribution modulation value; Activating the drilled hole differential pixel matrix through a probabilistic function to obtain a probabilistic drilled hole differential pixel matrix; After performing point subtraction between the probabilistic drilling differential pixel matrix and the drilling differential pixel distribution modulation value, taking the absolute value and calculating the negative of the logarithmic value with base 2 to obtain the probabilistic drilling differential pixel distribution modulation information matrix; After dividing the borehole differential pixel distribution characterization value by one minus the difference of each eigenvalue of the probabilistic borehole differential pixel matrix, summing all eigenvalues ​​of the probabilistic borehole differential pixel matrix and dividing by the scale of the borehole differential pixel matrix to obtain a probabilistic borehole differential pixel distribution modulation bias value; and The product of the probabilistic drilling differential pixel distribution modulation information matrix and the probabilistic drilling differential pixel distribution modulation bias value and the weight as a hyperparameter is multiplied to obtain an optimized drilling differential pixel matrix.

[0042] The optimization expression of the drilling differential pixel matrix is ​​as follows: in, is the probabilistic drilled differential pixel matrix, is the drilling differential pixel distribution characterization value, It is subtracted by position point. represents the logarithmic function value with base 2, It is added by location point. is the weight as a hyperparameter, is the dimension of the drilled differential pixel matrix, i.e., width times height, is each eigenvalue of the probabilistic drilled-difference pixel matrix, is the characteristic mean of the drilling difference pixel matrix, and are the maximum eigenvalue and the minimum eigenvalue of the drilling differential pixel matrix, respectively. is the optimized drilled differential pixel matrix.

[0043] That is, in the above preferred example, the probability information distribution planning based on pixel values ​​of the borehole differential pixel matrix is ​​carried out through the Bernoulli probability modulation distribution of the borehole differential pixel matrix relative to the pixel value distribution, and the probability reverse mapping of the probability characteristics of the borehole differential pixel matrix as a whole is used as an extended coverage of the set mapping space of the borehole differential pixel matrix to independently understand the interactive path between the intuitive probability information distribution and the abstract probability space mapping of the borehole differential pixel matrix, so as to improve the accuracy of the correction data obtained by the correction data generator based on the decoder input of the borehole differential pixel matrix by avoiding the counterfactual reasoning mapping of the outlier feature distribution of the borehole differential pixel matrix to the decoding regression probability. In this way, the image deformation (such as barrel distortion or pincushion distortion) caused by the optical characteristics of the sensor and the lens distortion can be corrected through machine vision technology and rotation correction technology, and 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 borehole image are extracted by using image recognition and analysis technology, which helps to more accurately identify and correct the deviation of the borehole and improve the processing accuracy.

[0044] It is worth mentioning that in other specific examples of the present application, the feature analysis based on the image foreground mask significant enhancement mechanism can be performed on the drilling image in other ways to generate correction data, for example: input the drilling image; use an image segmentation algorithm (such as U-Net or Mask R-CNN) to generate a foreground mask of the drilling image. The foreground mask is a binary image, in which the foreground area (drilling) is 1 and the background area is 0; the foreground mask is applied to the drilling image to enhance only the foreground area; the foreground area is enhanced using image enhancement technology (such as histogram equalization, sharpening or edge detection); the enhanced foreground area is combined with the original background area to generate correction data.

[0045] In particular, in S5, 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. In this way, the quality correction and control of the high-density circuit board is achieved to ensure the processing quality of the circuit board.

[0046] In summary, according to the embodiment of the present application, the method of high-density circuit board based on rapid heat dissipation of chips is explained, which uses image recognition and analysis technology based on machine vision to perform feature capture and significant feature enhancement processing of drilling images, so as to generate a drilling image after rotation correction, and realize the distortion and rotation correction of the drilling image. Then, the corrected drilling image is differentiated from the drilling reference image pixel by pixel to automatically generate the correction data. In this way, 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 it impossible to compare and correlate the two images, which helps to realize the quality correction and control of high-density circuit boards more intelligently, so as to ensure the processing quality of the circuit boards.

[0047] The embodiments of the present disclosure have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The selection of 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 persons of ordinary skill in the art 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.

2. The method for high-density circuit board based on rapid heat dissipation of chips according to claim 1, characterized in that: Performing feature analysis on the borehole image based on an image foreground mask significant enhancement mechanism to generate correction data, including: 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.

3. The method for high-density circuit board based on rapid heat dissipation of chips according to claim 2, 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.

4. The method for high-density circuit board based on rapid heat dissipation of chips according to claim 3, 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.

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 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.

6. The method for high-density circuit board based on rapid heat dissipation of chips according to claim 5, 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.

7. The method for high-density circuit board based on rapid heat dissipation of chips according to claim 6, 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.

8. The method for high-density circuit board based on rapid heat dissipation of chips according to claim 7, 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.

9. A high-density circuit board based on rapid heat dissipation of chips, characterized in that: The high-density circuit board device based on rapid chip heat dissipation is used to execute the high-density circuit board method based on rapid chip heat dissipation of claims 1-8.

Citation Information

Patent Citations

  • Method and system for correcting tool face of underground coal mine nearly-horizontal drilling image

    CN115661441A

  • PCB processing deviation correction method and system based on visual calibration

    CN116113164A

  • Quality control system, method and device for high-density circuit board and medium

    CN116847554A

  • High multi-layer precision plate production and processing system and method thereof

    CN117412521A

  • Artificial intelligence model method for target identification and detection

    CN117456300A