A process for high-density via hole deviation of ultra-thin substrates with laser drilling

Abstract

This application relates to the field of intelligent detection, and provides a process for detecting the hole deviation of high-density through-holes in an ultra-thin substrate. It uses image recognition and processing technologies based on computer vision to extract the drilling distribution features and enhance the grid significant features of the drilling distribution detection image and the drilling distribution design image, and thus intelligently obtains the hole deviation recognition result according to the drilling distribution difference between the enhanced features of the drilling distribution detection image and the drilling distribution design image. In this way, finer aperture details can be captured, the accuracy and precision of high-density through-hole hole deviation recognition are improved, the influence of human factors is reduced, the consistency and reliability of detection are ensured, and the recognition process becomes more automated and efficient.

Description

Technical Field

[0001] The present application relates to the field of intelligent detection, and more specifically, to a process for laser drilling high-density through holes on an ultra-thin substrate. Background Art

[0002] As electronic products develop towards smaller, thinner and lighter, higher requirements are placed on the design and manufacture of circuit boards. Ultra-thin substrates are one of the key technologies to achieve thin and light electronic devices. They not only help reduce the overall size and weight of the equipment, but also improve the integration and performance of electronic products. In the production of ultra-thin substrates, laser drilling is usually used to form through holes. However, due to the extremely thin thickness of the substrate, which is usually only tens of microns, and the smaller and smaller apertures, this high-density aperture is usually between tens of microns and hundreds of microns, which makes the position of the through holes easily shifted. This shift can lead to serious quality problems such as poor conduction, which not only affects the reliability of the product, but also greatly reduces the production yield. Therefore, the identification of hole deviation is crucial to ensure product quality and performance.

[0003] Traditional hole deviation identification methods usually rely on manual inspection or simple automated machine vision systems in ultra-thin substrate laser drilling high-density through-hole processes. However, manual inspection is easily affected by factors such as operator fatigue and lack of concentration, which will lead to a decrease in the consistency and accuracy of the inspection results, and cannot meet the dual requirements of speed and precision in modern high-density circuit board manufacturing. In addition, traditional machine vision systems may lack sufficient resolution and algorithm support in ultra-thin substrate laser drilling high-density through-hole processes, making it difficult to accurately identify tiny hole deviations. In other words, such systems usually rely on ordinary optical imaging equipment with limited resolution, which is insufficient to capture the details of tiny apertures on ultra-thin substrates, especially when the hole spacing is extremely small, resulting in inaccurate inspection results.

[0004] Therefore, an optimized process for laser drilling of high-density through holes on ultra-thin substrates is desired. Summary of the invention

[0005] In view of the shortcomings of the prior art, the present application provides a process for laser drilling of high-density through holes on an ultra-thin substrate.

[0006] According to one aspect of the present application, a process for radium drilling of high-density through holes on an ultra-thin substrate is provided, which comprises:

[0007] Provide ultra-thin substrate materials;

[0008] Cleaning the surface of the ultra-thin substrate material to obtain a cleaned ultra-thin substrate material;

[0009] Fix the cleaned ultra-thin substrate material on the processing table and use a vision system to accurately position the cleaned ultra-thin substrate material;

[0010] Use a high-precision laser device to perform laser drilling on the cleaned ultra-thin substrate material to obtain the drilled ultra-thin substrate material;

[0011] Use an optical detection device to perform hole deviation identification on the drilled ultra-thin substrate material to obtain a hole deviation identification result, and the hole deviation identification result is used to indicate whether there is a hole deviation.

[0012] Combined with the first aspect of the present application, in a process for hole deviation of high-density through holes in laser drilling of ultra-thin substrates in the first aspect of the present application, an optical detection device is used to perform hole deviation identification on the drilled ultra-thin substrate material to obtain a hole deviation identification result, and the hole deviation identification result is used to indicate whether there is a hole deviation, including: obtaining a drilled distribution detection image of the drilled ultra-thin substrate material collected by the optical detection device; extracting a drilled distribution design image from a database; inputting the drilled distribution detection image and the drilled distribution design image into a drilled distribution feature extractor to obtain a drilled distribution detection feature map and a drilled distribution design feature map; inputting the drilled distribution detection feature map and the drilled distribution design feature map into a feature gating enhancement module guided by grid energy saliency to obtain a grid granularity drilled distribution detection enhanced feature map and a grid granularity drilled distribution design enhanced feature map; calculating a drilled distribution difference feature map between the grid granularity drilled distribution detection enhanced feature map and the grid granularity drilled distribution design enhanced feature map; and obtaining the hole deviation identification result based on the drilled distribution difference feature map.

[0013] Due to the adoption of the above technical solutions, the present application has remarkable technical effects:

[0014] The process for hole deviation of high-density through holes in laser drilling of ultra-thin substrates provided by the present application uses computer vision-based image recognition and processing technologies to perform drilled distribution feature extraction and grid saliency feature enhancement on the drilled distribution detection image and the drilled distribution design image, and thus intelligently obtains the hole deviation identification result based on the drilled distribution difference between the features of the drilled distribution detection image and the drilled distribution design image after enhancement. In this way, finer hole diameter details can be captured, the accuracy and precision of high-density through hole hole deviation identification are improved, the influence of human factors is reduced, the consistency and reliability of detection are ensured, and the identification process is made more automated and efficient. Description of the Drawings

[0015] 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 drawings, the same reference numerals generally represent the same components or steps.

[0016] Figure 1 It is a flowchart of a process for hole deviation of high-density through-holes drilled by laser on an ultra-thin substrate according to an embodiment of the present application.

[0017] Figure 2 It is a flowchart of identifying hole deviation of the drilled ultra-thin substrate material using an optical detection device to obtain a hole deviation identification result in the process of hole deviation of high-density through-holes drilled by laser on an ultra-thin substrate according to an embodiment of the present application. The hole deviation identification result is used to indicate whether there is hole deviation.

[0018] Figure 3 It is a schematic diagram of data flow of identifying hole deviation of the drilled ultra-thin substrate material using an optical detection device to obtain a hole deviation identification result in the process of hole deviation of high-density through-holes drilled by laser on an ultra-thin substrate according to an embodiment of the present application. The hole deviation identification result is used to indicate whether there is hole deviation.

[0019] Figure 4 It is a flowchart of inputting the drilled hole distribution detection feature map and the drilled hole distribution design feature map into a feature gating enhancement module guided by grid energy saliency to obtain a grid granularity drilled hole distribution detection enhanced feature map and a grid granularity drilled hole distribution design enhanced feature map in the process of hole deviation of high-density through-holes drilled by laser on an ultra-thin substrate according to an embodiment of the present application. Detailed implementation manners

[0020] 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 the embodiments of the present application. It should be understood that the present application is not limited by the exemplary embodiments described herein.

[0021] In the development trend of electronic devices towards being thinner and more portable, the design and manufacturing of circuit boards face more stringent challenges. As one of the key technologies to achieve the thinness and lightness of devices, ultra-thin substrates play a significant role in improving the integration and performance of electronic products. In the manufacturing process of ultra-thin substrates, laser drilling technology is often used to form through-holes. Due to the thickness of the substrate usually being only dozens of micrometers, and the reduction of the hole diameter, the positioning accuracy of high-density through-holes becomes particularly important. Any slight hole position deviation may lead to a decrease in electrical conductivity, thereby affecting the reliability and production efficiency of the product.

[0022] Traditional hole deviation identification methods, whether relying on manual inspection or basic automated vision systems, have certain limitations. Manual inspection is vulnerable to operator fatigue and distraction, making it difficult to maintain consistency and accuracy in inspection, and unable to meet the requirements of speed and precision in high-density circuit board manufacturing. Traditional machine vision systems often struggle to accurately identify hole position deviations when dealing with tiny apertures on ultra-thin substrates due to insufficient resolution and algorithm support.

[0023] Based on this, the present application proposes a process for hole deviation of high-density through-holes in laser drilling of ultra-thin substrates. Figure 1 The flowchart of the process for hole deviation of high-density through-holes in laser drilling of ultra-thin substrates according to an embodiment of the present application is as follows. Figure 1 As shown, the process for hole deviation of high-density through-holes in laser drilling of ultra-thin substrates according to an embodiment of the present application includes: S110, providing an ultra-thin substrate material; S120, performing surface cleaning on the ultra-thin substrate material to obtain a cleaned ultra-thin substrate material; S130, fixing the cleaned ultra-thin substrate material on a processing table and using a vision system to precisely position the cleaned ultra-thin substrate material; S140, performing laser drilling on the cleaned ultra-thin substrate material using a high-precision laser device to obtain a laser-drilled ultra-thin substrate material; S150, using an optical detection device to identify hole deviation of the laser-drilled ultra-thin substrate material to obtain a hole deviation identification result, where the hole deviation identification result is used to indicate whether there is hole deviation.

[0024] In the process for hole deviation of high-density through-holes in laser drilling of ultra-thin substrates, first, an ultra-thin substrate material needs to be provided. The ultra-thin substrate material generally has an extremely thin thickness but sufficient strength and durability. Next, the surface of the ultra-thin substrate material needs to be cleaned to remove impurities, oil stains, and dust on the surface of the substrate material, which is beneficial to improving the positioning accuracy of the vision system and the quality of laser drilling, and avoiding processing errors caused by surface contaminants. Immediately afterwards, the cleaned ultra-thin substrate material is fixed on a processing table and precisely positioned using a vision system. It should be understood that fixing the ultra-thin substrate material can ensure stability during the processing, and precise positioning is a prerequisite for achieving high-density through-hole drilling. Any positioning deviation may lead to hole deviation, thereby affecting the electrical conductivity of the circuit board. Then, a high-precision laser device is used to perform laser drilling on the cleaned ultra-thin substrate material. High-density through-holes are formed on the ultra-thin substrate using laser drilling technology, and these through-holes are crucial for electrical connection between different layers of the circuit board. Finally, an optical detection device needs to be used to identify hole deviation of the laser-drilled ultra-thin substrate material. Hole deviation identification is a key step in quality control, which can timely detect and correct deviations during the drilling process, improving the yield and reliability of the product.

[0025] Accordingly, in the process of using an optical detection device to identify the hole deviation of the ultrathin substrate material after drilling to obtain a hole deviation identification result, the technical concept of this application is to acquire the drilling distribution detection image of the ultrathin substrate material after drilling collected by the optical detection device, extract the drilling distribution design image from the database, and use image recognition and processing techniques based on computer vision to extract the drilling distribution features and enhance the grid significant features of the drilling distribution detection image and the drilling distribution design image. Based on this, the hole deviation identification result is intelligently obtained according to the drilling distribution difference between the enhanced features of the drilling distribution detection image and the drilling distribution design image. In this way, finer hole diameter details can be captured, the accuracy and precision of high-density via hole deviation identification are improved, the influence of human factors is reduced, the consistency and reliability of detection are ensured, and the identification process becomes more automated and efficient.

[0026] Figure 2 It is a flowchart showing that in the process of the hole deviation of high-density through holes in an ultrathin substrate by laser drilling according to an embodiment of the present application, an optical detection device is used to identify the hole deviation of the ultrathin substrate material after drilling to obtain a hole deviation identification result, and the hole deviation identification result is used to indicate whether there is hole deviation. Figure 3 It is a schematic diagram of data flow showing that in the process of the hole deviation of high-density through holes in an ultrathin substrate by laser drilling according to an embodiment of the present application, an optical detection device is used to identify the hole deviation of the ultrathin substrate material after drilling to obtain a hole deviation identification result, and the hole deviation identification result is used to indicate whether there is hole deviation. As Figure 2 and Figure 3 shown, using an optical detection device to identify the hole deviation of the ultrathin substrate material after drilling to obtain a hole deviation identification result, and the hole deviation identification result is used to indicate whether there is hole deviation, including: S151, acquiring the drilling distribution detection image of the ultrathin substrate material after drilling collected by the optical detection device; S152, extracting the drilling distribution design image from the database; S153, inputting the drilling distribution detection image and the drilling distribution design image into a drilling distribution feature extractor to obtain a drilling distribution detection feature map and a drilling distribution design feature map; S154, inputting the drilling distribution detection feature map and the drilling distribution design feature map into a feature gating enhancement module guided by grid energy significance to obtain a grid granularity drilling distribution detection enhanced feature map and a grid granularity drilling distribution design enhanced feature map; S155, calculating the drilling distribution difference feature map between the grid granularity drilling distribution detection enhanced feature map and the grid granularity drilling distribution design enhanced feature map; S156, obtaining the hole deviation identification result based on the drilling distribution difference feature map.

[0027] In steps S151 and S152, obtain the drilling distribution detection image of the ultrathin substrate material after drilling collected by the optical detection device, and extract the drilling distribution design image from the database. It should be understood that the drilling distribution detection image contains information on the actual positions, sizes, and distribution of all the drill holes on the ultrathin substrate, and these information reflect the physical state after actual processing. The drilling distribution design image contains the drilling layout information expected in the design stage, specifically including the ideal positions, sizes, and shapes of the holes, the expected distances and arrangements between holes, etc. By automatically analyzing the drilling distribution detection image and the drilling distribution design image, the deviation between the actual drilling and the design can be identified, and the hole deviation identification result can be obtained, thereby greatly improving the detection speed and accuracy.

[0028] In step S153, input the drilling distribution detection image and the drilling distribution design image into a drilling distribution feature extractor to obtain a drilling distribution detection feature map and a drilling distribution design feature map. Specifically, in the embodiment of the present application, inputting the drilling distribution detection image and the drilling distribution design image into a drilling distribution feature extractor to obtain a drilling distribution detection feature map and a drilling distribution design feature map includes: inputting the drilling distribution detection image and the drilling distribution design image into a drilling distribution feature extractor based on a deep neural network model to obtain the drilling distribution detection feature map and the drilling distribution design feature map. The drilling distribution feature extractor based on a deep neural network model in the present application is a drilling distribution feature extractor based on an atrous convolutional neural network model. It should be understood that the drilling distribution detection image and the drilling distribution design image reflect the key feature information regarding the detection and design of the drilling distribution. Considering that the atrous convolution encoding method in the atrous convolutional neural network model can expand the receptive field without increasing the computational amount, which helps to capture the details of the micro-apertures on the ultrathin substrate and thus helps to identify whether there is hole deviation. Based on this, in the technical solution of the present application, input the drilling distribution detection image and the drilling distribution design image into a drilling distribution feature extractor based on an atrous convolutional neural network model to respectively capture and excavate the key drilling features in the detection image and the design image, and obtain a drilling distribution detection feature map and a drilling distribution design feature map.

[0029] In step S154, the drilling distribution detection feature map and the drilling distribution design feature map are input into a feature gating enhancement module guided by grid energy saliency to obtain a grid-grained drilling distribution detection enhanced feature map and a grid-grained drilling distribution design enhanced feature map. Correspondingly, considering that some regions or features in the drilling distribution detection feature map and the drilling distribution design feature map are more prominent or important than other parts, that is, the contribution degrees and saliencies of different local drilling regions to the whole are different. For example, holes located on key signal paths or connecting important components, the accuracy of their positions directly affects the functional performance of the circuit board. If these holes are offset, it may lead to signal transmission distortion or interruption, affecting the overall performance of the circuit board. Therefore, in order for the enhancement module to highlight the drilling features in important regions and suppress unimportant features, thereby improving the accuracy of hole deviation recognition. In the technical solution of this application, the drilling distribution detection feature map and the drilling distribution design feature map are input into a feature gating enhancement module guided by grid energy saliency to obtain a grid-grained drilling distribution detection enhanced feature map and a grid-grained drilling distribution design enhanced feature map. It is worth mentioning that the purpose of the feature gating enhancement module guided by grid energy saliency is to define grids to delimit the regions of interest and simulate the attention mechanism according to the principle of energy saliency, so as to enhance the expressiveness of the feature map and improve the overall performance of the model. That is, this method enables the model to focus more on the features that are more critical for task completion by identifying and emphasizing the key regions in the image while suppressing the relatively unimportant parts, thereby improving the recognition accuracy and processing efficiency.

[0030] Specifically, taking the drilling distribution detection feature map as an example, first, the drilling distribution detection feature map is divided into grids to obtain a set of local feature maps for drilling distribution detection. It should be understood that by dividing the feature map into multiple grids, the features of each local area for drilling distribution detection can be analyzed more meticulously, so that local details can be better captured, especially for those key areas that have a significant impact on the overall performance, such as high-density hole areas or holes on critical signal paths. Then, an energy significance descriptor for each local feature map is calculated to obtain a set of local energy significance descriptors for drilling distribution detection. In this way, it can reflect the feature intensity and importance within a certain area, helping the system distinguish which local features have a greater impact on the overall hole deviation detection result. Subsequently, the set of local energy significance descriptors is input into a local feature adaptive selector based on a gating function to obtain a set of local significant modulation weights for drilling distribution detection. That is, the gating function can selectively retain those features with higher significance while ignoring those features that have less impact on the detection result. And through the adaptive selector, it can ensure that the most critical features for hole deviation detection are processed preferentially. This can improve the detection accuracy, reduce false alarms and missed detections, and make hole deviation identification more accurate and reliable. Furthermore, using each significant modulation weight in the set of significant modulation weights as a weight, each local feature map in the set of local feature maps is weighted respectively to obtain a set of enhanced local feature maps. That is, the essence of the weighted enhancement step is attention application, which can make those local features with higher significance be given greater weights, thus being more prominent in the final feature map. Finally, the weighted feature maps are aggregated by the method of the above-mentioned grid division, and the aggregated features are input into a feature dispersion module based on a dilated convolutional layer to obtain a final grid-grained drilling distribution detection enhanced feature map. In particular, since the module mainly focuses on grid-level details, there may be a problem of unnatural feature transition at the edges of the grid. In order to achieve smooth diffusion of features, the aggregated feature map is processed by applying a dilated convolutional layer to reduce the abrupt change in feature expression at the grid edges.

[0031] Specifically, Figure 4 In the process of ultra-thin substrate laser drilling of high-density through holes according to an embodiment of the present application, the flowchart of inputting the drilling distribution detection feature map and the drilling distribution design feature map into a feature gating enhancement module guided by grid energy significance to obtain a grid-grained drilling distribution detection enhanced feature map and a grid-grained drilling distribution design enhanced feature map is as follows. As Figure 4As shown, input the drilling distribution detection feature map and the drilling distribution design feature map into the feature gating enhancement module guided by grid energy saliency to obtain the grid-granularity drilling distribution detection enhanced feature map and the grid-granularity drilling distribution design enhanced feature map, including: S1541, perform grid division on the drilling distribution detection feature map to obtain a set of drilling distribution detection local feature maps; S1542, calculate the energy saliency description factors of each drilling distribution detection local feature map in the set of drilling distribution detection local feature maps to obtain a set of drilling distribution detection local energy saliency description factors; S1543, input the set of drilling distribution detection local energy saliency description factors into the local feature adaptive selector based on the gating function to obtain a set of drilling distribution detection local significant modulation weights; S1544, use each drilling distribution detection local significant modulation weight in the set of drilling distribution detection local significant modulation weights as weights to respectively weight each drilling distribution detection local feature map in the set of drilling distribution detection local feature maps to obtain a set of drilling distribution detection local enhanced feature maps; S1545, perform feature aggregation on the set of drilling distribution detection local enhanced feature maps in the manner of the grid division to obtain a drilling distribution detection local significant guidance enhanced feature map; S1546, input the drilling distribution detection local significant guidance enhanced feature map into the feature dispersion module based on the dilated convolutional layer to obtain the grid-granularity drilling distribution detection enhanced feature map.

[0032] More specifically, in the embodiment of the present application, calculating the energy saliency description factors of each drilling distribution detection local feature map in the set of drilling distribution detection local feature maps to obtain a set of drilling distribution detection local energy saliency description factors includes: respectively extracting the maximum value and the minimum value of the drilling distribution detection local feature map to obtain the drilling distribution detection local feature maximum value and the drilling distribution detection local feature minimum value; calculating the difference between the drilling distribution detection local feature maximum value and the drilling distribution detection local feature minimum value to obtain the drilling distribution detection local feature difference value; respectively calculating the global mean and variance of the drilling distribution detection local feature map to obtain the drilling distribution detection local feature mean and the drilling distribution detection local feature variance; adding the value obtained by multiplying the drilling distribution detection local feature variance by the constant two and the drilling distribution detection local feature mean with the hyperparameter to obtain the drilling distribution detection local feature statistical modulation value; dividing the drilling distribution detection local feature difference value by the drilling distribution detection local feature statistical modulation value to obtain the drilling distribution detection local energy saliency description factor corresponding to the drilling distribution detection local feature map.

[0033] More specifically, in the embodiments of the present application, inputting the set of local energy significance description factors for borehole distribution detection into a local feature adaptive selector based on a gating function to obtain a set of local significant modulation weights for borehole distribution detection includes: normalizing the set of local energy significance description factors for borehole distribution detection to obtain a set of normalized local energy significance description factors for borehole distribution detection; inputting each normalized local energy significance description factor in the set of normalized local energy significance description factors for borehole distribution detection into a gating function for masking processing to obtain the set of local significant modulation weights for borehole distribution detection.

[0034] More specifically, in the embodiments of the present application, normalizing the set of local energy significance description factors for borehole distribution detection to obtain a set of normalized local energy significance description factors for borehole distribution detection includes: calculating an exponential function with the negative of each local energy significance description factor in the set of local energy significance description factors for borehole distribution detection as the exponent and the natural constant e as the base to obtain a set of local energy significance description exponential factors for borehole distribution detection; calculating the reciprocal of the sum of each local energy significance description exponential factor in the set of local energy significance description exponential factors for borehole distribution detection and the constant one to obtain the set of normalized local energy significance description factors for borehole distribution detection.

[0035] In the embodiments of the present application, specifically, inputting the borehole distribution detection feature map and the borehole distribution design feature map into a feature gating enhancement module guided by grid energy significance to obtain a grid-grained borehole distribution detection enhanced feature map and a grid-grained borehole distribution design enhanced feature map includes: inputting the borehole distribution detection feature map into a feature gating enhancement module guided by grid energy significance and processing it according to the following significant gating enhancement formula to obtain the grid-grained borehole distribution detection enhanced feature map; where the significant gating enhancement formula is: Where is the borehole distribution detection feature map, is a grid division operation, are respectively each borehole distribution detection local feature map in the set of borehole distribution detection local feature maps, is the th borehole distribution detection local feature map in the set of borehole distribution detection local feature maps, and respectively extract the maximum and minimum values in the feature map, and are respectively 's mean and variance, is a hyperparameter, denotes the drilling distribution detection local energy significance description factor corresponding to the th local feature map of drilling distribution detection, is the masking process, is the th local significant modulation weight of drilling distribution detection corresponding to the local feature map of drilling distribution detection, is a predetermined threshold, is to perform a feature aggregation operation on each weighted feature map, is the local significant guidance enhanced feature map of the drilling distribution detection, is the dilated convolution operation, is the enhanced feature map of grid granularity drilling distribution detection. Similarly, the encoding method of the drilling distribution design feature map is also as shown in the above formula.

[0036] In step S155, calculate the drilling distribution difference feature map between the enhanced feature map of grid granularity drilling distribution detection and the enhanced feature map of grid granularity drilling distribution design. Specifically, in the embodiment of the present application, calculating the drilling distribution difference feature map between the enhanced feature map of grid granularity drilling distribution detection and the enhanced feature map of grid granularity drilling distribution design includes: calculating the position-wise difference between the enhanced feature map of grid granularity drilling distribution detection and the enhanced feature map of grid granularity drilling distribution design to obtain the drilling distribution difference feature map. It should be understood that in order to identify and quantify the deviation between the actual drilling result and the design expectation, so as to highlight the difference between the actual drilling and the design, including position offset, size difference or shape deviation, etc., in the technical solution of the present application, calculate the position-wise difference between the enhanced feature map of grid granularity drilling distribution detection and the enhanced feature map of grid granularity drilling distribution design to obtain the drilling distribution difference feature map.

[0037] In step S156, based on the drilling distribution difference feature map, obtain the hole deviation identification result. Specifically, in the embodiment of the present application, obtaining the hole deviation identification result based on the drilling distribution difference feature map includes: inputting the drilling distribution difference feature map into a hole deviation identifier based on a classifier to obtain the hole deviation identification result. That is, perform classification processing on the drilling distribution difference feature map obtained by performing differential calculation between the enhanced feature map of grid granularity drilling distribution detection and the enhanced feature map of grid granularity drilling distribution design, so as to intelligently obtain the hole deviation identification result. In this way, finer aperture details can be captured, the accuracy and precision of high-density via hole deviation identification are improved, the influence of human factors is reduced, the consistency and reliability of detection are ensured, and the identification process is made more automated and efficient.

[0038] In a preferred example, since the grid granularity drilling distribution detection enhanced feature map and the grid granularity drilling distribution design enhanced feature map respectively represent the local spatial enhanced image semantic features of the drilling distribution detection image and the drilling distribution design image based on the grid distribution energy significance of the image semantic features, when calculating the position-by-position difference therebetween to obtain the drilling distribution difference feature map, the drilling distribution difference feature map will also have an offset in the aggregation of the image semantic feature differences caused by the inconsistency of the local spatial energy enhancement of the image semantic features under the source image semantic distribution. Therefore, it is desired to improve the comprehensibility of the aggregation of the image semantic feature differences of the drilling distribution difference feature map, thereby improving the accuracy of the hole deviation recognition result obtained by inputting the drilling distribution difference feature map into the classifier-based hole deviation recognizer.

[0039] Therefore, when the present application inputs the drilling distribution difference feature map into the classifier-based hole deviation recognizer, the drilling distribution difference feature map is optimized, including the steps of: expanding the drilling distribution difference feature map into a drilling distribution difference feature vector; subtracting the 0-norm of the drilling distribution difference feature vector from the length of the drilling distribution difference feature vector to obtain a drilling distribution difference isolated representation value; calculating the base-2 logarithm of the sum of the square of the drilling distribution difference isolated representation value and the drilling distribution difference isolated representation value to obtain a drilling distribution difference information order value; calculating a power function with each eigenvalue of the drilling distribution difference feature vector as the base and the difference between the drilling distribution difference isolated representation value minus one as the exponent, and performing a dot product with the drilling distribution difference information order value to obtain a drilling distribution difference pre-guidance vector; performing a dot product of the drilling distribution difference feature vector with the difference between the drilling distribution difference isolated representation value minus one, and then performing a dot product with the reciprocal of the drilling distribution difference isolated representation value to obtain a drilling distribution difference field constraint vector; calculating an exponential function with the natural constant as the base and each eigenvalue of the drilling distribution difference field constraint vector as the exponent to obtain a drilling distribution difference field bias vector; and performing a dot addition of the drilling distribution difference pre-guidance vector and the drilling distribution difference field bias vector to obtain an optimized drilling distribution difference feature vector.

[0040] Among them, the optimization process of the drilling distribution difference feature vector is expressed as: Among them, represents the drilling distribution difference feature vector, represents the length of the drilling distribution difference feature vector, represents the 0-norm of the vector, represents the drilling distribution difference isolated representation value, represents dot product by position, represents the logarithm with base 2, represents dot addition by position, represents the natural exponential function, is to calculate a power function with each eigenvalue of the differential feature vector of the drilling distribution as the base and the difference between the differential isolation representation value of the drilling distribution minus one as the exponent, represents the optimized differential feature vector of the drilling distribution.

[0041] Thus, for the high-dimensional feature manifold of the differential feature map of the drilling distribution, the vector field representation with the eigenvalues of the feature set as the aggregation dimension is used. The superimposed value of the vector field of the differential feature vector of the drilling distribution after unfolding the differential feature map of the drilling distribution at the isolated zero position is used as the order information to fix the local position of the eigenvalues of its feature set, and a bias for the reversibility of the feature regression distribution field of the differential feature vector of the drilling distribution is added as a reward to achieve the mapping target tracking of the regression distribution of the differential feature vector of the drilling distribution to the eigenvalue position, so that the feature set of the differential feature map of the drilling distribution perceives the mapping migration to the aggregation distribution, thereby improving the understandability of the differential aggregation mapping of the image semantic features of the differential feature map of the drilling distribution and improving the accuracy of the hole deviation recognition result obtained by inputting the hole deviation recognizer based on the classifier. In this way, finer aperture details can be captured, the accuracy and precision of high-density via hole deviation recognition are improved, the influence of human factors is reduced, the consistency and reliability of detection are ensured, and the recognition process becomes more automated and efficient.

[0042] In summary, the process of hole deviation of high-density through holes in ultra-thin substrates based on the embodiments of the present application is clarified. It uses image recognition and processing techniques based on computer vision to extract the drilling distribution features and enhance the grid significant features of the drilling distribution detection image and the drilling distribution design image, and then intelligently obtains the hole deviation recognition result according to the drilling distribution difference between the enhanced features of the drilling distribution detection image and the drilling distribution design image. In this way, finer aperture details can be captured, the accuracy and precision of high-density through hole deviation recognition are improved, the influence of human factors is reduced, the consistency and reliability of detection are ensured, and the recognition process becomes more automated and efficient.

[0043] The basic principles of the present application have been described above in conjunction with specific embodiments. However, it should be noted that the advantages, advantages, effects, etc. mentioned in the present application are only examples and not limitations. It cannot be considered that these advantages, advantages, effects, etc. are essential for each embodiment of the present application. In addition, the specific details of the above application are only for the purpose of illustration and easy understanding, rather than limitations. The above details do not limit the present application to necessarily adopt the above specific details to implement.

Claims

1. A process for laser drilling high-density through-holes on ultra-thin substrates, characterized in that: include: Provide ultra-thin substrate materials; Cleaning the surface of the ultra-thin substrate material to obtain a cleaned ultra-thin substrate material; Fixing the cleaned ultra-thin substrate material on a processing table, and using a visual system to accurately position the cleaned ultra-thin substrate material; Using high-precision laser equipment to perform laser drilling on the cleaned ultra-thin substrate material to obtain a drilled ultra-thin substrate material; Using an optical detection device to identify the hole deviation of the ultra-thin substrate material after drilling to obtain a hole deviation identification result, wherein the hole deviation identification result is used to indicate whether there is a hole deviation; Wherein, an optical detection device is used to identify the hole deviation of the ultra-thin substrate material after drilling to obtain a hole deviation identification result, and the hole deviation identification result is used to indicate whether there is a hole deviation, including: Acquire a drilling distribution detection image of the ultra-thin substrate material after drilling collected by the optical detection device; extracting a drilling distribution design image from a database; Inputting the drilling distribution detection image and the drilling distribution design image into a drilling distribution feature extractor to obtain a drilling distribution detection feature map and a drilling distribution design feature map; Inputting the drilling distribution detection feature map and the drilling distribution design feature map into a feature gating enhancement module guided by grid energy significance to obtain a grid-size drilling distribution detection enhancement feature map and a grid-size drilling distribution design enhancement feature map; Calculating a drilling distribution differential feature map between the grid granularity drilling distribution detection enhancement feature map and the grid granularity drilling distribution design enhancement feature map; Based on the borehole distribution differential feature map, obtaining the borehole deviation identification result; The drilling distribution detection feature map and the drilling distribution design feature map are input into a feature gating enhancement module based on grid energy significance guidance to obtain a grid-size drilling distribution detection enhancement feature map and a grid-size drilling distribution design enhancement feature map, including: Performing grid division on the borehole distribution detection feature map to obtain a set of local feature maps of borehole distribution detection; Calculating the energy significance description factor of each local feature map of borehole distribution detection in the set of local feature maps of borehole distribution detection to obtain a set of local energy significance description factors of borehole distribution detection; Inputting the set of local energy significance descriptors for drilling distribution detection into a local feature adaptive selector based on a gating function to obtain a set of local significant modulation weights for drilling distribution detection; Taking each local significant modulation weight for drilling distribution detection in the set of local significant modulation weights for drilling distribution detection as a weight, weighting each local feature map for drilling distribution detection in the set of local feature maps for drilling distribution detection respectively to obtain a set of local enhanced feature maps for drilling distribution detection; Performing feature aggregation on the set of the local enhanced feature maps for borehole distribution detection according to the grid division method to obtain a local significant guided enhanced feature map for borehole distribution detection; The drilling distribution detection local significant guided enhanced feature map is input into the feature dispersion module based on the hole convolution layer to obtain the grid-granularity drilling distribution detection enhanced feature map.

2. The process for radium drilling high-density through-holes on ultra-thin substrates according to claim 1, characterized in that: The drilling distribution detection image and the drilling distribution design image are input into a drilling distribution feature extractor to obtain a drilling distribution detection feature map and a drilling distribution design feature map, including: inputting the drilling distribution detection image and the drilling distribution design image into a drilling distribution feature extractor based on a deep neural network model to obtain the drilling distribution detection feature map and the drilling distribution design feature map.

3. The process for radium drilling high-density through-holes on ultra-thin substrates according to claim 2 is characterized in that: The drilling distribution feature extractor based on the deep neural network model is a drilling distribution feature extractor based on the void convolutional neural network model.

4. The process for radium drilling of high-density through holes on ultra-thin substrates according to claim 3 is characterized in that: Calculating the energy significance description factor of each local feature map of the borehole distribution detection in the set of local feature maps of the borehole distribution detection to obtain a set of local energy significance description factors of the borehole distribution detection, including: Respectively extracting the maximum value and the minimum value of the local feature map of the drilling distribution detection to obtain the local feature maximum value and the local feature minimum value of the drilling distribution detection; Calculating the difference between the maximum value of the local feature of the drilling distribution detection and the minimum value of the local feature of the drilling distribution detection to obtain a local feature difference value of the drilling distribution detection; Respectively calculating the global mean and variance of the local feature map of the borehole distribution detection to obtain the local feature mean of the borehole distribution detection and the local feature variance of the borehole distribution detection; The value obtained by multiplying the local feature variance of the borehole distribution detection by a constant of two and the mean value of the local feature of the borehole distribution detection are added to the hyperparameter to obtain a local feature statistical modulation value of the borehole distribution detection; The local feature difference value of the drilling distribution detection is divided by the local feature statistical modulation value of the drilling distribution detection to obtain a local energy significance description factor of the drilling distribution detection corresponding to the local feature map of the drilling distribution detection.

5. The process for radium drilling of high-density through holes on ultra-thin substrates according to claim 4 is characterized in that: The set of local energy significance descriptors for drilling distribution detection is input into a local feature adaptive selector based on a gating function to obtain a set of local significant modulation weights for drilling distribution detection, including: Normalizing the set of local energy significance description factors for borehole distribution detection to obtain a set of normalized local energy significance description factors for borehole distribution detection; Each normalized local energy significance description factor of drilling distribution detection in the set of normalized local energy significance description factors of drilling distribution detection is input into a gating function for masking processing to obtain a set of local significance modulation weights of drilling distribution detection.

6. The process for radium drilling of high-density through holes on ultra-thin substrates according to claim 5, characterized in that: The set of local energy significance description factors for borehole distribution detection is normalized to obtain a set of normalized local energy significance description factors for borehole distribution detection, including: Calculate an exponential function with the natural constant e as the base and the negative number of each borehole distribution detection local energy significance description factor in the set of borehole distribution detection local energy significance description factors as the exponent to obtain a set of borehole distribution detection local energy significance description exponential factors; The reciprocal of the sum of each borehole distribution detection local energy significance description index factor in the set of borehole distribution detection local energy significance description index factors and a constant one is calculated to obtain the set of normalized borehole distribution detection local energy significance description factors.

7. The process for radium drilling of high-density through holes on ultra-thin substrates according to claim 6, characterized in that: Calculating a drilling distribution differential feature map between the grid granularity drilling distribution detection enhanced feature map and the grid granularity drilling distribution design enhanced feature map, including: calculating the positional difference between the grid granularity drilling distribution detection enhanced feature map and the grid granularity drilling distribution design enhanced feature map to obtain the drilling distribution differential feature map.

8. The process for radium drilling of high-density through holes on ultra-thin substrates according to claim 7, characterized in that: Based on the borehole distribution differential feature map, the borehole deviation identification result is obtained, including: inputting the borehole distribution differential feature map into a classifier-based borehole deviation identifier to obtain the borehole deviation identification result.

Images