Image recognition method and system applied to chip layout

Through image recognition methods, the empty bridge layout feature mining and spatial feature state analysis are solved, and the traditional chip layout analysis is achieved, efficient and accurate chip layout analysis is achieved, and design quality and performance are improved.

CN118968539AActive Publication Date: 2024-11-15CHENGDU DAHAN MICROWAVE INTEGRATION TECH CO LTD
View PDF 9 Cites 0 Cited by

Patent Information

Application Number
CN202411166769.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-23
Publication Date
2024-11-15
Estimated Expiration
2044-08-23

AI Technical Summary

Technical Problem

Traditional chip layout analysis methods are inefficient and difficult to ensure the accuracy and comprehensiveness of the analysis. Especially when dealing with complex chip layouts, it is impossible to effectively identify the empty bridge layout features and the state conflict weights between quantization areas.

Method used

Provide an image recognition method, by obtaining the target chip layout, performing empty bridge layout feature mining and spatial feature state analysis, generating empty bridge layout feature label data, characterizing empty bridge layout features in regional layout image binary, and calculating the status Conflict weights, optimized analysis using wavelet vector and residual wavelet vector mining.

Benefits of technology

It realizes efficient and accurate chip layout analysis, provides comprehensive design reference, improves chip design quality and performance, shortens design cycles, and reduces design costs.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118968539A_ABST
    Figure CN118968539A_ABST
Patent Text Reader

Abstract

The embodiment of the invention relates to the technical field of image processing, in particular to an image recognition method and system applied to a chip layout, and successfully solves many problems existing in a traditional chip layout analysis method through innovative empty bridge layout feature mining operation and an accurate spatial feature state analysis method. In addition, the embodiment of the invention further provides abundant air bridge layout feature annotation data, and provides more comprehensive and deep design reference for designers. Through a series of processing and analysis steps, the brand-new and efficient chip layout analysis method is provided, the overall quality and performance of chip design can be improved, the design period can be shortened to a great extent, the design cost can be reduced to a great extent, and therefore remarkable benefit improvement is brought to the whole chip design industry.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of image processing technology, and in particular to an image recognition method and system applied to chip layout. Background Art

[0002] In the field of chip design, accurate spatial feature analysis of chip layout has always been one of the core challenges of the industry. Traditional chip layout analysis methods often rely on the experience and manual operation of designers, which is not only inefficient but also difficult to ensure the accuracy and comprehensiveness of the analysis. Especially when dealing with complex chip layouts, the limitations of traditional methods are more prominent. It is often impossible to effectively identify the layout features of empty bridges in published drawings, and it is even more difficult to quantify the state conflict weights between different areas.

[0003] In order to solve the above problems, the industry has been seeking a method that can automatically and efficiently analyze the spatial characteristics of chip layouts. However, due to the complexity and diversity of chip layouts, the realization of this goal faces many technical difficulties. Summary of the invention

[0004] In order to at least overcome the above-mentioned deficiencies in the prior art, one of the objectives of the present application is to provide an image recognition method and system for chip layout.

[0005] The present application embodiment provides an image recognition method applied to a chip layout, which is applied to an image recognition system. The method includes: Acquire a target chip layout, wherein the target chip layout includes a first region layout image and a second region layout image; Performing an empty bridge layout feature mining operation on the target chip layout to obtain a mining result, wherein the empty bridge layout feature mining operation is used to extract empty bridge layout features from a regional layout image binary consisting of the first regional layout image and the second regional layout image, wherein the empty bridge layout feature includes a chip skeleton line block binary consisting of a first chip skeleton line block of the first regional layout image and a second chip skeleton line block of the second regional layout image, and the spatial feature commonality scores of the first chip skeleton line block and the second chip skeleton line block meet a set commonality discrimination condition; On the basis that the mining result indicates that there are empty bridge layout features in the regional map image binary, generating empty bridge layout feature annotation data based on the mining result; The target chip layout is spatially characterized in state by analyzing the spatial feature state according to the air bridge layout feature annotation data to obtain a spatial feature state analysis viewpoint, wherein the spatial feature state analysis viewpoint represents a state conflict weight between the first area layout image and the second area layout image.

[0006] Optionally, performing an empty bridge layout feature mining operation on the target chip layout to obtain a mining result includes: Extracting the empty bridge layout features in a preset image data set, wherein the preset image data set includes at least one associated chip skeleton line block binary; If there is a target-associated chip skeleton line map block binary in the preset image data set, a mining result including the target-associated chip skeleton line map block binary is obtained, and the target-associated chip skeleton line map block binary includes the first chip skeleton line map block and the second chip skeleton line map block; wherein each of the associated chip skeleton line map block binary matches the following conditions at the same time: the associated chip skeleton line map block binary includes a main chip main skeleton line map block and a target chip skeleton line map block; the main chip main skeleton line map block and the target chip skeleton line map block correspond to a main area layout image and a target area layout image from an associated area layout image binary, respectively; the associated area layout image binary is identified as a non-associated area layout image in the first spatial feature state parsing network; The main chip main skeleton line map block in the main area layout image is updated to the target chip skeleton line map block to obtain an updated associated area layout image binary group, and the updated associated area layout image binary group is identified as an associated area layout image in the first spatial feature state parsing network.

[0007] Optionally, the mining result includes at least one chip skeleton line block binary, and the generating of the empty bridge layout feature annotation data based on the mining result includes: generating a block coding feature vector according to each chip skeleton line block binary; generating a first state parsing instruction, the first state parsing instruction is used to characterize that the second space feature state parsing network performs a space feature state parsing project according to the block coding feature vector; generating the empty bridge layout feature annotation data based on the block coding feature vector and the first state parsing instruction; The method of performing spatial feature state analysis on the target chip layout based on the empty bridge layout feature annotation data to obtain the spatial feature state analysis viewpoint includes: entering the empty bridge layout feature annotation data and the target chip layout into the second spatial feature state analysis network to obtain the spatial feature state analysis viewpoint.

[0008] Optionally, the method further includes: On the basis that the mining result indicates that there is no empty bridge layout feature in the target chip layout, a second state parsing instruction is generated, wherein the second state parsing instruction is used to indicate that the second space feature state parsing network directly executes the space feature state parsing project; The second state analysis instruction and the target chip layout are entered into the second spatial feature state analysis network to obtain the spatial feature state analysis viewpoint.

[0009] Optionally, the method includes: the second spatial feature state parsing network performs the following processing: Performing a wavelet vector mining operation on a chip layout set to obtain a first chip layout wavelet vector, wherein the chip layout set is a set obtained by combining target instruction data, the first area layout image, and the second area layout image, wherein the target instruction data includes the empty bridge layout feature labeling data or the second state parsing instruction; Performing a residual wavelet vector mining operation on the first chip layout wavelet vector to obtain a second chip layout wavelet vector; The target subvector of the second chip layout wavelet vector is subjected to spatial feature state mining to obtain the spatial feature state analysis viewpoint, wherein the spatial feature state analysis viewpoint represents the spatial feature commonality score of the first area layout image and the second area layout image.

[0010] Optionally, the second spatial feature state parsing network is debugged based on the following ideas: Acquire a historical chip layout, wherein the historical chip layout includes a first historical region layout image, a second historical region layout image, and training annotations, wherein the training annotations represent state conflict between the first historical region layout image and the second historical region layout image; Performing an empty bridge layout feature mining operation on the historical chip layout to obtain a historical mining result; If the historical mining result indicates that there is no empty bridge layout feature in the historical chip layout, the second state parsing instruction is determined as historical instruction data; if the historical mining result indicates that there is an empty bridge layout feature in the historical chip layout, historical empty bridge layout feature annotation data is generated; the historical instruction data is determined according to the historical empty bridge layout feature annotation data; Entering the historical instruction data and the historical chip layout into the second spatial feature state analysis network to obtain a historical spatial feature state analysis viewpoint; Based on the difference between the historical spatial feature state parsing views and the training annotations, the neural network weights of the second spatial feature state parsing network are optimized.

[0011] Optionally, the method further includes: Acquire multiple positive samples, in which the main historical area map image and the target historical area map image have state conflict; Entering each of the positive examples into the first spatial feature state analysis network to obtain a corresponding spatial feature state analysis viewpoint, and determining the positive examples that do not have state conflict represented by the spatial feature state analysis viewpoint as target positive examples; For each of the target positive examples, the first target chip skeleton line map block in the main historical area layout image in the target positive example is updated to the second target chip skeleton line map block in the target historical area layout image to obtain a reference main historical area layout image; if the spatial feature state analysis viewpoint obtained by entering the reference main historical area layout image and the target historical area layout image into the first spatial feature state analysis network represents that there is a state conflict, the first target chip skeleton line map block and the second target chip skeleton line map block are determined as an associated chip skeleton line map block tuple, and the associated chip skeleton line map block tuple is added to the preset image data set; Among them, the first target chip skeleton line map block is any chip skeleton line map block of the main historical area layout image, and the second target chip skeleton line map block is any chip skeleton line map block of the target historical area layout image.

[0012] Optionally, the method further includes: The target chip layout is adjusted based on the spatial feature state analysis viewpoint.

[0013] Optionally, the method further includes: The target chip layout is stored based on the spatial feature state analysis viewpoint.

[0014] An embodiment of the present application also provides an image recognition system, including a processor and a memory and a bus connected to the processor; wherein the processor and the memory communicate with each other through the bus; the processor is used to call program instructions in the memory to execute the above-mentioned image recognition method applied to chip layout.

[0015] An embodiment of the present application also provides a computer-readable storage medium on which a program is stored, and when the program is executed by a processor, the above-mentioned image recognition method applied to chip layout is implemented.

[0016] Beneficial Effects The image recognition method and system applied to chip layout provided in the embodiment of the present application successfully solves many problems existing in the traditional chip layout analysis method through innovative empty bridge layout feature mining operation and accurate spatial feature state analysis method. Not only that, the embodiment of the present application further provides rich empty bridge layout feature annotation data, providing designers with a more comprehensive and in-depth design reference. Through a series of processing and analysis steps, a new and efficient chip layout analysis method is provided, which not only helps to improve the overall quality and performance of chip design, but also greatly shortens the design cycle and reduces design costs, thereby bringing significant benefits to the entire chip design industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0018] Figure 1 A flowchart of an image recognition method applied to chip layout provided in an embodiment of the present application.

[0019] Figure 2 A block diagram of an image recognition system provided in an embodiment of the present application.

[0020] icon: 100- Image recognition system; 101 - processor; 102 - memory; 103 - bus. DETAILED DESCRIPTION

[0021] The exemplary embodiments disclosed in the present application will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present application are shown in the accompanying drawings, it should be understood that the present application can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present application and to fully convey the scope of the present application to those skilled in the art.

[0022] In order to better understand the above technical scheme, the technical scheme of the present application is described in detail below through the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present application and the specific features in the embodiments are detailed descriptions of the technical scheme of the present application, rather than limitations on the technical scheme of the present application. In the absence of conflict, the embodiments of the present application and the technical features in the embodiments can be combined with each other.

[0023] Figure 1 This is a flowchart of an image recognition method applied to a chip layout according to an embodiment of the present application, which is applied to an image recognition system and includes steps 110 to 140.

[0024] Step 110: Acquire a target chip layout, where the target chip layout includes a first region layout image and a second region layout image.

[0025] In the embodiment of the present application, the target chip layout is a physical graphic representation converted by designers from the circuit schematic, which accurately depicts key information such as the layout, connection, and size of each component on the chip. The target chip layout refers to the specific chip layout to be analyzed and processed by the image recognition system, which reflects the final physical implementation of the chip design. In this application scenario, the target chip layout is a complex and precise graphic that contains detailed information such as circuit components, wiring, and pads, and is the direct basis for manufacturing chips.

[0026] In the target chip layout, the layout can be divided into different functional areas for easy design and management. The first area layout image refers to a specific area image with certain functions or characteristics in the target chip layout. This area can contain specific circuit modules, logic units or wiring structures, and is an important part of the chip function realization. In the application scenario, the first area layout image and the second area layout image cooperate with each other to form a complete chip design. Similar to the first area layout image, the second area layout image is also an important part of the target chip layout. It represents another area in the chip with a specific function or layout. The two areas can differ in function and layout, but they are both key parts for the normal operation of the chip. The second area layout image can contain detailed information such as different circuit elements, wiring structures, and connections with other areas, which is crucial to ensuring the overall performance and functionality of the chip.

[0027] In the process of chip design, obtaining an accurate and complete target chip layout is a crucial step. The image recognition system first connects with the interface of the chip design software to automatically obtain the latest target chip layout data. These data not only include the overall layout of the chip, but also describe the detailed information of each area in detail. Specifically, when the image recognition system executes step 110, it will export the required layout files from the chip design software. These files can be stored in a specific format (such as GDSII or OASIS) and contain all the physical layer information and hierarchical structure of the chip design. The system parses these files and extracts the complete data of the target chip layout. During the extraction process, the system pays special attention to two important areas in the layout: the first area layout image and the second area layout image. These two areas play different roles in chip design and can be responsible for different functions or logical operations respectively. Therefore, the system needs to accurately identify and divide these two areas so that they can be analyzed and processed more deeply later. In order to ensure the accurate division of the area, the image recognition system uses advanced image processing technology and algorithms to perform edge detection, area segmentation and other operations on the layout. Through these processes, the system can accurately locate the boundary between the first area map image and the second area map image, laying a solid foundation for subsequent feature mining and spatial state analysis.

[0028] Step 120, perform an empty bridge layout feature mining operation on the target chip layout to obtain a mining result, wherein the empty bridge layout feature mining operation is used to extract empty bridge layout features in the regional layout image tuple composed of the first regional layout image and the second regional layout image, and the empty bridge layout features include a chip skeleton line map block tuple composed of a first chip skeleton line map block of the first regional layout image and a second chip skeleton line map block of the second regional layout image, and the spatial feature commonality scores of the first chip skeleton line map block and the second chip skeleton line map block meet the set commonality judgment conditions.

[0029] In an embodiment of the present application, the empty bridge layout feature mining operation is a specific data processing process that aims to identify and extract features related to the empty bridge layout from the target chip layout. The empty bridge is a wire structure used to connect different levels or areas in chip design, which is crucial for the functional realization of the chip. The empty bridge layout feature mining operation searches and locates these empty bridge structures in the first area layout image and the second area layout image through image processing and pattern recognition technology, and then analyzes its layout, size, connection method and other key features.

[0030] The mining results refer to the data and information obtained after the feature mining operation of the empty bridge layout. These results describe in detail the distribution, type, size and other related parameters of the empty bridges in the chip layout. The accuracy and completeness of the mining results are crucial for the subsequent spatial feature state analysis, and they provide strong data support for the optimization and improvement of chip design.

[0031] In the chip layout, skeleton blocks refer to graphic elements that depict the main structure and connection relationship of the chip. The first chip skeleton blocks specifically refer to such graphic elements extracted from the first area layout image, which represent the key circuit structure and connection path in the area. These blocks are of great significance for understanding the function and design intent of the chip. Similar to the first chip skeleton blocks, the second chip skeleton blocks are key graphic elements extracted from the second area layout image. They reflect the main structure and connection method of the circuit in the area and are an important part of chip design. By comparing and analyzing the skeleton blocks of the two areas, we can gain a deeper understanding of the relationship and differences between them.

[0032] The spatial feature commonality score is a quantitative indicator used to evaluate the similarity or consistency of the spatial layout of the first chip skeleton line block and the second chip skeleton line block. The higher the score, the more commonality the two blocks have in spatial features, that is, they are more similar in layout, size, shape, etc. This score helps to identify the layout rules or design patterns of air bridges that can exist in the chip layout.

[0033] The commonality judgment condition is a preset standard or threshold used to determine whether the spatial feature commonality score of two chip skeleton line blocks reaches a certain level. When the score exceeds this set condition, the system considers that the two blocks have significant commonality in spatial layout and can belong to similar air bridge layout modes. This set condition can be adjusted according to actual needs to adapt to different application scenarios and design requirements.

[0034] When executing step 120, the image recognition system will conduct an in-depth analysis of the target chip layout to explore the key features of the air bridge layout. The core goal of this step is to find and extract feature information related to the air bridge layout in the first area layout image and the second area layout image. The system will first use advanced image processing algorithms to carefully process and analyze the layout images of the two areas. This includes steps such as edge detection, morphological processing, and feature extraction, aiming to extract key air bridge layout features from complex image data. Then, the system identifies and locates the first chip skeleton line block and the second chip skeleton line block in the processed image. These blocks are the core structure of chip design and are crucial to understanding the rules of air bridge layout. The system compares and analyzes the spatial features of these blocks, such as position, size, shape, etc., to explore the commonalities and differences between them. In order to quantify the similarity of these spatial features, the system calculates the spatial feature commonality score of the first chip skeleton line block and the second chip skeleton line block. This score reflects the similarity of the two blocks in spatial layout, and the higher the score, the closer their spatial features are. Finally, the system evaluates these scores based on a set commonality discrimination condition. When the spatial feature commonality score of a chip skeleton line block pair meets this condition, the system considers that the pair of blocks have significant similarities in the layout of air bridges. This similarity can reveal a certain law or pattern in chip design, which has important guiding significance for subsequent optimization and improvement. Through this series of operations and analysis, the image recognition system can accurately mine the air bridge layout features in the target chip layout, providing strong data support for the subsequent spatial feature state analysis.

[0035] Step 130: On the basis that the mining result indicates that there are empty bridge layout features in the regional map image binary, empty bridge layout feature annotation data is generated based on the mining result.

[0036] In the embodiment of the present application, the air bridge layout feature annotation data refers to the detailed annotation information generated according to the mining results after the air bridge layout feature mining operation. These data not only accurately mark the position of the air bridge in the figure, but also include the size, shape, type and other key parameters related to the air bridge layout. The air bridge layout feature annotation data are crucial for the subsequent spatial feature state analysis, chip design verification and optimization, because they provide accurate and comprehensive information about the air bridge layout.

[0037] On the basis that the mining results clearly indicate that there are empty bridge layout features in the regional layout image binary (i.e., the first regional layout image and the second regional layout image), the image recognition system further processes these data to generate annotation data for the empty bridge layout features. First, the system carefully reviews the mining results to ensure that the identified empty bridge layout features are accurate and reliable. This includes verifying key information such as the location, size, and shape of the empty bridge to ensure the accuracy of these data. Then, the system uses professional annotation tools and techniques to make detailed annotations for the mined empty bridge layout features. In this process, the system generates a complete set of annotation data for each empty bridge, which not only includes the precise location of the empty bridge in the image, but also covers key information such as its size, type, and connection method. In addition, in order to enhance the readability and usability of the annotation data, the system also uses a standardized format to store and display this data. This means that whether it is designers, engineers or other relevant personnel, they can easily understand and use these annotation data to better verify, optimize and analyze chip design. Finally, the generated empty bridge layout feature annotation data will be properly saved by the system and can be seamlessly connected with other design tools and platforms to achieve data sharing and collaborative work. This not only improves the efficiency of chip design, but also greatly enhances the accuracy and reliability of the design. It can be seen that step 130 is a key link to ensure that the air bridge layout features are accurately marked, which lays a solid foundation for the subsequent spatial feature state analysis and further optimization of chip design.

[0038] Step 140: perform spatial feature state analysis on the target chip layout according to the air bridge layout feature annotation data to obtain a spatial feature state analysis viewpoint, wherein the spatial feature state analysis viewpoint represents a state conflict weight between the first area layout image and the second area layout image.

[0039] In the embodiment of the present application, spatial feature state analysis is a method for in-depth analysis of the chip layout, which focuses on studying and understanding the spatial relationship and feature state between different areas in the layout. In this process, the system uses the feature annotation data of the air bridge layout, combined with the rules and principles of layout design, to perform a detailed state interpretation and relationship combing of each element in the layout. In this way, the potential connections, conflicts or synergies between different areas can be revealed, thereby providing strong support for the optimization of chip design.

[0040] Spatial feature state analysis viewpoints refer to professional insights or conclusions formed during the spatial feature state analysis process. These viewpoints can be based on in-depth data analysis and understanding of layout design, and can reveal the state relationship between different regions in the publication map, such as whether there is conflict, synergy or independence. These viewpoints are of great significance in guiding the improvement and optimization of chip design.

[0041] The state conflict weight is a quantitative indicator used to evaluate the degree of state conflict between different regions in the layout. In chip design, different regions may have state conflicts due to functional requirements, routing restrictions and other factors. These conflicts may affect the performance and stability of the chip. The state conflict weight is a numerical value used to measure the severity of these conflicts. It can help designers better understand the potential problems in the layout and formulate corresponding optimization strategies.

[0042] In step 140, the image recognition system will use the previously generated empty bridge layout feature annotation data to perform a more in-depth spatial feature state analysis of the target chip layout. First, the system loads the empty bridge layout feature annotation data, which records in detail the key information such as the position, size, and type of the empty bridge in the layout. Next, the system analyzes the first area layout image and the second area layout image one by one based on this data, combined with the professional knowledge and rules of chip design. During the analysis process, the system pays special attention to the spatial relationship between the two areas. For example, the system checks whether there is a layout conflict between the empty bridges in the two areas, such as overlapping positions, mismatched sizes, etc. At the same time, the system also analyzes the connection relationship between the empty bridges to determine whether there is a potential circuit connection problem. In order to quantify these conflicts and potential problems, the system calculates an important indicator: the state conflict weight. This weight value can intuitively reflect the severity of the state conflict between the two areas. The higher the weight value, the more serious the conflict, and the designer needs to pay more attention and optimization. Finally, the system integrates all the analysis results into a comprehensive spatial feature state analysis view. This view not only includes the specific status and problems of the empty bridge layout in the layout, but also provides a quantitative evaluation of the state conflict weight. Designers can optimize and improve chip design in a targeted manner based on these viewpoints to improve chip performance and stability. It can be seen that step 140 is a process of in-depth analysis, quantitative evaluation and forming professional viewpoints, which provides strong data support and direction guidance for further optimization of chip design.

[0043] Combining the above content, in a complete application scenario, the image recognition system is used to analyze a complex chip layout. This chip layout consists of two main areas, namely the first area layout image and the second area layout image. These two areas are different in function and layout, but they are interrelated, so accurate analysis of them is crucial to ensure the correctness of chip design.

[0044] First, in step 110, the image recognition system automatically obtains the complete layout of the target chip. This step is achieved by interfacing with the chip design software to ensure the accuracy and up-to-dateness of the layout data. The obtained layout data is accurately divided into the first area layout image and the second area layout image by the system for subsequent feature mining.

[0045] Next, in step 120, the image recognition system performs an empty bridge layout feature mining operation on the target chip layout. The core of this operation is to extract the features of the empty bridge layout in the layout images of the first area and the second area. The system uses advanced image processing technology and machine learning algorithms to identify the chip skeleton line blocks in the two areas and construct a chip skeleton line block binary. The spatial feature commonality scores between these blocks meet a preset commonality judgment condition, which means that they have significant similarities in spatial layout, which is crucial for judging the rationality of the empty bridge layout and optimizing the layout.

[0046] When the mining results confirm that there are empty bridge layout features in the regional layout image binary, step 130 is executed. In this step, the image recognition system generates annotation data of the empty bridge layout features based on the mining results. These data mark the location, size and other key parameters of the empty bridge in detail, providing accurate data support for the subsequent spatial feature state analysis.

[0047] Finally, in step 140, the image recognition system uses the generated empty bridge layout feature annotation data to perform in-depth spatial feature state analysis of the target chip layout. Through complex algorithm analysis, the system derives a spatial feature state analysis viewpoint, which specifically characterizes the state conflict weight between the first area layout image and the second area layout image. This weight value is crucial for evaluating the rationality and performance of chip design. It can help designers optimize layout, reduce potential conflicts, and thus improve the overall performance of the chip.

[0048] Through this specific application scenario, the chip layout can be accurately acquired and analyzed, key air bridge layout features can be extracted, and finally a spatial feature state analysis viewpoint that helps optimize the design can be generated. This series of automated processing flows not only improves the efficiency of chip design, but also greatly improves the accuracy and reliability of the design.

[0049] The embodiment of the present application achieves efficient and accurate analysis of the internal spatial layout of the chip by deeply mining the empty bridge layout features in the target chip layout. Specifically, through the comprehensive processing of the first area layout image and the second area layout image, the embodiment of the present application not only extracts the key empty bridge layout features, but also further generates detailed empty bridge layout feature annotation data. These data provide a solid foundation for the subsequent spatial feature state analysis, allowing designers to more clearly understand the spatial relationship between the various elements inside the chip.

[0050] It is worth mentioning that the embodiment of the present application can accurately identify the chip skeleton line blocks with commonality by calculating the spatial feature commonality scores of the first chip skeleton line block and the second chip skeleton line block and setting commonality discrimination conditions. This innovative method not only improves the accuracy of the recognition of the air bridge layout features, but also effectively reduces the misjudgment rate, providing strong support for the optimization of chip design.

[0051] Furthermore, the embodiment of the present application performs an in-depth spatial feature state analysis of the target chip layout based on the generated empty bridge layout feature annotation data. Through this step, the designer can intuitively understand the state conflict weight between the first area layout image and the second area layout image, which is crucial for discovering and resolving potential design conflicts.

[0052] In summary, the embodiment of the present application provides a new and efficient chip layout analysis method for chip designers through a series of processing and analysis steps. This not only helps to improve the overall quality and performance of chip design, but also greatly shortens the design cycle and reduces design costs, thereby bringing significant benefits to the entire chip design industry.

[0053] In some optional embodiments, the target chip layout is subjected to an empty bridge layout feature mining operation to obtain a mining result, including: extracting empty bridge layout features in a preset image data set, the preset image data set including at least one associated chip skeleton line map block binary; if a target associated chip skeleton line map block binary exists in the preset image data set, then obtaining a mining result including the target associated chip skeleton line map block binary, the target associated chip skeleton line map block binary including the first chip skeleton line map block and the second chip skeleton line map block; wherein each of the associated chip skeleton line map block binary simultaneously matches the following conditions: the associated chip skeleton line map block binary The block binary group includes a main chip main skeleton line map block and a target chip skeleton line map block; the main chip main skeleton line map block and the target chip skeleton line map block respectively correspond to the main area layout image and the target area layout image from the associated area layout image binary group; the associated area layout image binary group is identified as a non-associated area layout image in the first spatial feature state parsing network; the main chip main skeleton line map block in the main area layout image is updated to the target chip skeleton line map block to obtain an updated associated area layout image binary group, and the updated associated area layout image binary group is identified as an associated area layout image in the first spatial feature state parsing network.

[0054] In this embodiment, first, the image recognition system extracts the empty bridge layout features from a preset image data set. The preset image data set contains multiple associated chip skeleton line block tuples, which have been processed and analyzed by the system before and have certain representativeness and learning value.

[0055] Next, the system checks whether there is a target-associated chip skeleton line image block binary in the preset image data set. If so, the binary, including the first chip skeleton line image block and the second chip skeleton line image block, will be used as part of the mining result.

[0056] However, this associated chip skeleton line block binary is not randomly selected, and it needs to meet a series of conditions. Specifically, each associated chip skeleton line block binary must meet the following conditions at the same time: the binary must include a main chip main skeleton line block and a target chip skeleton line block. These two blocks come from the main area layout image and the target area layout image in the associated area layout image binary. This associated area layout image binary was initially identified as a non-associated area layout image in a network called the first spatial feature state parsing. This means that in the initial state, the two areas are not considered to be associated with each other. However, when the system updates the main chip main skeleton line block in the main area layout image to the target chip skeleton line block, this updated associated area layout image binary is identified as an associated area layout image in the first spatial feature state parsing network. This change shows that by updating the blocks, the system can discover that there is actually some connection between the seemingly unrelated areas.

[0057] The associated chip skeleton line block pairs that meet these conditions not only reveal the potential empty bridge layout features in the chip layout, but also provide valuable information for further learning and optimization of the system.

[0058] Through the detailed explanation of the above embodiments, the empty bridge layout features can be accurately mined in complex chip layouts through deep learning and pattern recognition technology. The application of this method not only improves the efficiency and accuracy of chip design, but also provides designers with a new perspective and tool to understand and optimize their designs. In this way, through innovative image recognition technology and deep learning methods, the automatic mining and annotation of empty bridge layout features in chip layouts are realized. This not only greatly reduces the workload of manual analysis, but also improves the accuracy and comprehensiveness of the analysis. At the same time, through the spatial feature state analysis viewpoint generated by the mining results, designers can more intuitively understand the spatial relationship between the elements inside the chip, so as to promptly discover and solve potential design problems. Therefore, this technical solution has brought significant beneficial effects to the chip design industry, including improving design efficiency, reducing costs, and optimizing product performance.

[0059] In other optional embodiments, the mining results include at least one chip skeleton line block binary, and the generation of empty bridge layout feature annotation data based on the mining results includes: generating a block coding feature vector according to each chip skeleton line block binary; generating a first state parsing instruction, the first state parsing instruction is used to characterize that the second spatial feature state parsing network executes a spatial feature state parsing project according to the block coding feature vector; and generating the empty bridge layout feature annotation data based on the block coding feature vector and the first state parsing instruction.

[0060] And the spatial feature state analysis of the target chip layout is performed based on the empty bridge layout feature annotation data to obtain the spatial feature state analysis viewpoint, including: entering the empty bridge layout feature annotation data and the target chip layout into the second spatial feature state analysis network to obtain the spatial feature state analysis viewpoint.

[0061] In some other optional embodiments, during the mining and annotation process of the empty bridge layout features by the image recognition system, the mining results include at least one chip skeleton line block binary. These binary groups reflect the potential associations and features between different areas in the chip layout. Next, the system will generate empty bridge layout feature annotation data based on these mining results.

[0062] First, the image recognition system generates a corresponding block encoding feature vector for each chip skeleton line block binary. These feature vectors are digital representations of the chip skeleton line blocks. They capture the key features of the blocks, such as shape, size, positional relationship, etc., and encode this information into a numerical form that can be processed by computers.

[0063] Then, the system generates the first state parsing instruction. This instruction is designed specifically to guide the second spatial feature state parsing network. It tells the network how to perform the spatial feature state parsing project based on the previously generated tile encoding feature vector. In simple terms, this instruction is the "guide" for the network to perform parsing work.

[0064] With the block encoding feature vector and the first state analysis instruction, the image recognition system can generate the empty bridge layout feature annotation data. These data not only contain the key feature information of the chip skeleton line block, but also incorporate the instructions on how to analyze these features, providing comprehensive data support for the subsequent spatial feature state analysis.

[0065] The next step is to analyze the spatial feature state of the target chip layout based on the empty bridge layout feature annotation data. Figure 1 The data is then entered into the second spatial feature state analysis network. This network is specially designed to analyze the spatial feature state in the chip layout. It can conduct in-depth analysis and interpretation of the layout based on the features and instructions in the annotated data.

[0066] Through network processing, the system will eventually output a spatial feature state analysis view. This view is a comprehensive interpretation of the spatial relationship between different areas in the target chip layout. It reveals the state conflict weights between elements in the layout and provides valuable optimization suggestions for chip designers.

[0067] It can be seen that by generating block encoding feature vectors and state analysis instructions, the empty bridge layout feature annotation data is generated, and these data are used to conduct in-depth spatial feature state analysis of the target chip layout. The application of this method not only improves the intelligent level of chip design, but also provides designers with more accurate and efficient optimization tools.

[0068] In detail, the image recognition system automatically generates the feature annotation data of the empty bridge layout, and the spatial feature state analysis is performed based on this data, thereby obtaining a comprehensive spatial feature state analysis viewpoint. This greatly improves the automation and design efficiency of chip design, reduces the workload of manual analysis, and ensures the accuracy and objectivity of the analysis. In addition, through the state conflict weights revealed in the analysis viewpoint, designers can optimize the chip layout design more specifically, thereby improving the overall performance and stability of the chip. Therefore, this technical solution has brought significant beneficial effects to the field of chip design.

[0069] In some alternative embodiments, the method further includes: generating a second state analysis instruction based on the fact that the mining result characterizes that there is no empty bridge layout feature in the target chip layout, the second state analysis instruction being used to characterize that the second spatial feature state analysis network directly executes the spatial feature state analysis project; entering the second state analysis instruction and the target chip layout into the second spatial feature state analysis network to obtain the spatial feature state analysis viewpoint.

[0070] Based on this embodiment, the method executed by the image recognition system further includes an additional step, which is for processing situations where there are no obvious empty bridge layout features in the target chip layout.

[0071] Specifically, when the mining results show that there are no obvious empty bridge layout features in the target chip layout, the system generates a second state parsing instruction. This instruction is different from the first state parsing instruction mentioned above. It no longer relies on a specific tile encoding feature vector for parsing, but directly guides the second spatial feature state parsing network to perform the spatial feature state parsing project.

[0072] In other words, the second state parsing instruction instructs the second spatial feature state parsing network: "In this chip layout, there is no obvious empty bridge layout feature for reference, so please directly perform spatial feature state parsing on the entire layout." The image recognition system then uses this second state parsing instruction and the target chip version Figure 1 After receiving the instructions and layout data, the network will perform a comprehensive spatial feature state analysis on the entire layout according to the instructions.

[0073] This analysis process can involve a comprehensive analysis of the position, size, shape and other characteristics of each element in the layout, as well as the relationship and influence between these elements. Through complex calculations and reasoning of the network, the system will eventually output a spatial feature state analysis view.

[0074] This view is also an interpretation of the spatial relationship in the target chip layout, but it is obtained without obvious air bridge layout features. It can rely more on other significant spatial features in the layout, such as the layout of components and the direction of connections, to reveal the state conflict weights between elements in the layout.

[0075] In this way, the image recognition system can handle chip layouts that do not have obvious empty bridge layout features. The system generates a second state analysis instruction and directly guides the second spatial feature state analysis network to perform a comprehensive spatial feature state analysis, thereby obtaining a comprehensive analysis viewpoint. In detail, this embodiment enhances the flexibility and adaptability of the image recognition system. Even in the face of chip layouts without obvious empty bridge layout features, the system can provide designers with valuable optimization suggestions through comprehensive spatial feature state analysis. This not only improves the efficiency and quality of the design, but also reduces the dependence on specific features, making the system more widely applicable.

[0076] Under some optional design ideas, the method includes: the second spatial feature state analysis network implements the following processing: performing a wavelet vector mining operation on a chip layout set to obtain a first chip layout wavelet vector, the chip layout set is a set obtained by combining target instruction data, the first area layout image and the second area layout image, and the target instruction data includes the empty bridge layout feature annotation data or the second state analysis instruction; performing a residual wavelet vector mining operation on the first chip layout wavelet vector to obtain a second chip layout wavelet vector; performing spatial feature state mining on the target subvector of the second chip layout wavelet vector to obtain the spatial feature state analysis viewpoint, and the spatial feature state analysis viewpoint represents the spatial feature commonality score of the first area layout image and the second area layout image.

[0077] Based on this design idea, the image recognition system adopts a specific method to process the chip layout data and obtain the spatial feature state analysis point of view.

[0078] First, the image recognition system builds a chip layout set. This set is obtained by combining the target instruction data, the first area layout image, and the second area layout image. The target instruction data plays a key role here. It can be the previously mentioned empty bridge layout feature annotation data or the second state parsing instruction, depending on the previous processing steps and the type of data obtained.

[0079] Next, the system uses the second spatial feature state parsing network to perform wavelet vector mining on this chip layout set. Wavelet vector mining is an effective data analysis method that can extract key features from the data. Through this operation, the system obtains the first chip layout wavelet vector. These wavelet vectors capture the important features in the chip layout set and provide a basis for subsequent analysis.

[0080] The system then performs a residual wavelet vector mining operation on the first chip layout wavelet vector. The purpose of this step is to further refine the features. By calculating the residual (i.e., the difference between the actual value and the predicted value), the system can more accurately capture the subtle changes in the chip layout. This operation will produce a second chip layout wavelet vector, which more accurately represents the characteristics of the chip layout.

[0081] Finally, the system performs spatial feature state mining on the target subvector of the second chip layout wavelet vector. This step is to reveal the spatial feature commonality between the first region layout image and the second region layout image by performing in-depth analysis on specific wavelet vectors. The system evaluates the similarity of the two regions in spatial features and gives a spatial feature commonality score. This score reflects the degree of coordination between the two regions in layout and design, which is crucial for chip design optimization.

[0082] Through the above steps, the image recognition system can comprehensively analyze the chip layout data and obtain accurate spatial feature state analysis views. This method not only improves the accuracy of data analysis, but also provides valuable feedback information for chip designers, helping them to find and solve potential problems in the design stage. In this way, through wavelet vector mining and residual wavelet vector mining operations, the characteristics of chip layout data are deeply analyzed, and accurate spatial feature state analysis views are obtained. This not only improves the intelligent level of chip design, but also provides designers with strong data support, helping them optimize design solutions and improve chip performance and reliability.

[0083] It is worth mentioning that in the above design ideas, wavelet vector mining operations and residual wavelet vector mining operations constitute the core technical links, which are crucial to improving the accuracy and efficiency of the image recognition system.

[0084] In simple terms, the wavelet vector mining operation is to use wavelet transform technology to conduct in-depth analysis of the chip layout set. Wavelet transform, as a powerful mathematical tool, can capture detailed information of signals at multiple scales and directions. In image processing, it is particularly suitable for extracting local features of images, such as edges, textures, etc. By performing wavelet vector mining on the chip layout set, the system can generate the first chip layout wavelet vectors, which not only contain the overall information of the layout, but also accurately reflect the detailed features of each component in the layout.

[0085] The residual wavelet vector mining operation further refines information based on wavelet vector mining. In image processing, residual usually refers to the difference between the actual image and the processed or transformed image. By performing residual analysis on the first chip layout wavelet vector, the system can capture important features that may be ignored or weakened during the processing. This residual wavelet vector mining not only enhances the sensitivity of the features, but also helps to discover small changes and anomalies in the image, thereby improving the accuracy of image recognition.

[0086] In summary, wavelet vector mining and residual wavelet vector mining complement each other and together constitute the core technology of the image recognition system. The former fully extracts image features through wavelet transform, while the latter further refines these features through residual analysis, ensuring that the system can accurately and efficiently identify and analyze complex structures and states in the chip layout. The combined application of these two technologies significantly improves the performance and reliability of the image recognition system when processing complex image data.

[0087] In some preferred embodiments, the second spatial feature state analysis network is debugged based on the following ideas: obtaining a historical chip layout, the historical chip layout including a first historical area layout image, a second historical area layout image and training annotations, the training annotations characterizing the state conflict between the first historical area layout image and the second historical area layout image; performing an empty bridge layout feature mining operation on the historical chip layout to obtain a historical mining result; if the historical mining result characterizes that there is no empty bridge layout feature in the historical chip layout, determining the second state analysis instruction as historical instruction data; if the historical mining result characterizes that there is an empty bridge layout feature in the historical chip layout, generating historical empty bridge layout feature annotation data; determining the historical instruction data based on the historical empty bridge layout feature annotation data; entering the historical instruction data and the historical chip layout into the second spatial feature state analysis network to obtain a historical spatial feature state analysis viewpoint; optimizing the neural network weights of the second spatial feature state analysis network based on the difference between the historical spatial feature state analysis viewpoint and the training annotations.

[0088] Based on this embodiment, the debugging process of the image recognition system for the second spatial feature state analysis network is very rigorous and meticulous. This process not only involves in-depth analysis of historical chip layouts, but also includes optimization of network weights to improve the accuracy and performance of the system.

[0089] First, the system acquires a series of historical chip layouts. These layouts consist of a first historical region layout image, a second historical region layout image, and corresponding training annotations. The training annotations are critical because they clearly indicate the state conflicts between the first historical region layout image and the second historical region layout image. This conflict may be caused by design mismatches, layout errors, or functional conflicts, and the training annotations provide the system with criteria for judging these conflicts.

[0090] Next, the system mines the empty bridge layout features of these historical chip layouts. The purpose of this step is to find out whether there are empty bridge layout features in the layout, because these features may affect the performance of the chip. The mining operation carefully analyzes every detail in the layout to obtain accurate results.

[0091] After obtaining the mining results, the system performs a series of judgments and operations. If the mining results show that there is no empty bridge layout feature in the historical chip layout, the system directly determines the second state parsing instruction as the historical instruction data. This means that in this case, the system does not need additional empty bridge layout feature annotation data, but can directly perform subsequent spatial feature state parsing.

[0092] However, if the mining results indicate that there are empty bridge layout features in the historical chip layout, the system will generate corresponding historical empty bridge layout feature annotation data. These data are a detailed description of the empty bridge layout features in the layout, which is crucial for the subsequent spatial feature state analysis.

[0093] Once the historical instruction data is determined (whether based on the air bridge layout feature annotation data or directly determined), the system will combine these instruction data with the historical chip version. Figure 1 The data is then entered into the second spatial feature state analysis network. Through this step, the system can obtain a historical spatial feature state analysis view. This view is based on the network's understanding and analysis of the current input data, and it reflects the spatial characteristics and relationships between various regions in the map.

[0094] Finally, the system optimizes the neural network weights of the second spatial feature state parsing network based on the difference between the historical spatial feature state parsing viewpoints and the training annotations. This process is done by comparing the network output with the actual training annotations, with the goal of allowing the network to better learn and understand the spatial feature states in the layout. By continuously adjusting and optimizing the weights, the system can gradually improve the accuracy of the parsing, thereby providing more reliable support for subsequent chip design.

[0095] It can be seen that the second spatial feature state parsing network is optimized through a sophisticated debugging process, so that it can more accurately parse the spatial feature states in the chip layout. This not only improves the performance of the system, but also provides chip designers with more accurate and valuable feedback information, helping them to discover and solve potential problems in the design stage. The beneficial effect of this debugging method is that it improves the accuracy and reliability of the image recognition system in the field of chip layout analysis, and provides strong technical support for the optimization and innovation of chip design.

[0096] In some optional embodiments, the method further includes: obtaining multiple positive samples, in which the main historical area layout image and the target historical area layout image have state conflict; entering each of the positive samples into the first spatial feature state analysis network to obtain the corresponding spatial feature state analysis viewpoint, and determining the positive samples that do not have state conflict according to the spatial feature state analysis viewpoint as target positive samples; for each of the target positive samples, updating the first target chip skeleton line map block in the main historical area layout image in the target positive sample to the second target chip skeleton line map block in the target historical area layout image to obtain the reference main historical area. Domain map image; if the spatial feature state analysis viewpoint representation obtained by entering the reference main historical area map image and the target historical area map image into the first spatial feature state analysis network has state conflict, the first target chip skeleton line map block and the second target chip skeleton line map block are determined as an associated chip skeleton line map block tuple, and the associated chip skeleton line map block tuple is added to the preset image data set; wherein, the first target chip skeleton line map block is any chip skeleton line map block of the main historical area map image, and the second target chip skeleton line map block is any chip skeleton line map block of the target historical area map image.

[0097] Based on this embodiment, the image recognition system adopts a unique method to enrich and improve the preset image data set so as to more accurately identify and analyze the state conflicts in the chip layout. The following are the detailed implementation steps of the method: First, the system obtains multiple positive samples. These positive samples are specially selected, and each sample contains a main historical area layout image and a target historical area layout image, and these two parts of the image have conflicting states. This conflict may be caused by differences in layout design, incompatibility between circuit components, or other performance issues caused by other factors.

[0098] Next, the system enters each positive example into the first spatial feature state parsing network for analysis. This network is specially trained to recognize and parse the spatial feature states in the layout image. Through the network's processing, the system obtains a corresponding spatial feature state parsing view, which clearly indicates whether there is a state conflict in the image.

[0099] Then, the system screens out those positive examples that are interpreted as not having state conflicts and identifies them as target positive examples. The purpose of this step is to find those examples that were not found to have conflicts in the initial analysis, but may actually hide deeper conflicts or potential problems.

[0100] For each target positive example, the system performs a refined analysis step. Specifically, the system replaces the first target chip skeleton line block in the main historical area layout image with the second target chip skeleton line block in the target historical area layout image to obtain a reference main historical area layout image. This replacement operation is performed while keeping all other elements unchanged in order to more accurately locate the blocks that may cause conflicts.

[0101] After the replacement is completed, the system will refer to the main historical area layout image and the target historical area layout image and enter the first spatial feature state analysis network again for analysis. If the result of this analysis shows that there is a state conflict between the two, then the system can determine that the previously replaced pair of chip skeleton line blocks (i.e., the first target chip skeleton line block and the second target chip skeleton line block) are related and may cause a conflict.

[0102] Finally, the system adds this pair of associated chip skeleton line tiles as a binary to the preset image dataset. The advantage of this is that by continuously accumulating and enriching such associated conflicting tile pairs, the image recognition system can gradually improve its ability to identify and prevent similar conflicts.

[0103] In this way, by deeply mining and analyzing the potential conflicts in the positive samples, the system not only improves the recognition accuracy of state conflicts, but also provides valuable reference information for subsequent optimization design. With the continuous improvement and expansion of the preset image data set, the performance and accuracy of the image recognition system in processing complex chip layouts will also be significantly improved.

[0104] In some other possible embodiments, the method further includes: adjusting the target chip layout based on the spatial feature state analysis viewpoint.

[0105] In detail, this step is performed after the system has obtained the spatial feature state analysis view of the chip layout through the previous processing flow. First, the spatial feature state analysis view is the conclusion drawn by the image recognition system after an in-depth analysis of the chip layout, which describes in detail the spatial features of each part of the layout and the relationship between them. This view not only points out possible problems or conflicts in the layout, but also provides valuable information on how to optimize the layout structure. Next, the image recognition system uses this spatial feature state analysis view as a guide to adjust the target chip layout. The purpose of the adjustment is mainly to resolve the conflicts in the layout and optimize its spatial layout to improve the performance and reliability of the chip. During the adjustment process, the system may involve repositioning, resizing or replacing certain elements in the chip layout. For example, if it is found that there is too close contact between two adjacent circuit elements, which may cause heat accumulation or signal interference, the system will appropriately increase the distance between them or add an isolation structure according to the recommendations of the spatial feature state analysis view. In addition, the image recognition system may also optimize certain areas in the layout as a whole based on the commonality scores mentioned in the spatial feature state analysis view. For example, if the commonality score of a certain area is low, it means that there may be a lot of room for improvement in the spatial layout of the area. The system can improve its overall performance and stability by adjusting the layout, connection method or power supply network of the circuit components in the area. After the adjustment is completed, the image recognition system will once again perform spatial feature state analysis on the adjusted chip layout to ensure that all adjustments have achieved the expected results and no new problems have been introduced. In general, by adjusting the target chip layout based on the viewpoint of spatial feature state analysis, the image recognition system can effectively solve the problems and conflicts in the layout, optimize its spatial layout, and thus improve the overall performance and reliability of the chip. This adjustment method is not only highly targeted and flexible, but also can perform personalized optimization processing according to actual needs, providing strong technical support for chip design.

[0106] In some other possible embodiments, the method further includes: storing the target chip layout based on the spatial feature state analysis viewpoint.

[0107] Based on this embodiment, the method adopted by the image recognition system further expands its functions, including storing the target chip layout based on the spatial feature state analysis viewpoint. This step is performed after the chip layout is deeply analyzed and the corresponding spatial feature state analysis viewpoint is obtained.

[0108] First, when the image recognition system completes the spatial feature state analysis of the target chip layout, it will generate a detailed spatial feature state analysis view. This view not only reveals the spatial relationship between the elements in the layout, but also points out possible optimization points or potential problems. Next, the system uses this formed spatial feature state analysis view to decide how to store the target chip layout more efficiently. The storage process is not simply saving the layout data to the hard disk or database, but optimizing it according to the analysis view to ensure that the stored layout data is both complete and efficient. During the storage process, the image recognition system may compress or encode the layout data according to the suggestions of the spatial feature state analysis view to reduce the storage space requirements while maintaining the integrity and accessibility of the data. For example, the system may identify repeated or similar patterns in the publication and represent them in a more compact way to save storage space. In addition, the storage method based on the spatial feature state analysis view also helps to improve the efficiency of subsequent retrieval and analysis of the chip layout. Because the features of the layout data have been extracted and optimized during storage, the system can locate key information more quickly when needed, reducing processing time. Finally, this storage method based on the perspective of spatial feature state analysis also provides a valuable resource library for the chip design team. Designers can easily access optimized layout data, thereby accelerating the design and development process of new chips. In summary, by storing the target chip layout based on the perspective of spatial feature state analysis, the image recognition system not only optimizes the storage efficiency, but also facilitates subsequent chip design and analysis. This method not only saves storage space and improves data processing efficiency, but also brings more efficient workflows and richer resource support to the chip design industry.

[0109] An embodiment of the present application provides a computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements the image recognition method applied to chip layout.

[0110] An embodiment of the present application provides a processor, which is used to run a program, wherein the image recognition method applied to chip layout is executed when the program is run.

[0111] In the present application embodiment, Figure 2 As shown, the image recognition system 100 includes at least one processor 101, and at least one memory 102 and a bus 103 connected to the processor 101; wherein the processor 101 and the memory 102 communicate with each other through the bus 103; the processor 101 is used to call the program instructions in the memory 102 to execute the above-mentioned image recognition method applied to the chip layout.

[0112] The present application is described with reference to the flowcharts and / or block diagrams of the methods, image recognition systems (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0113] In a typical configuration, the image recognition system includes one or more processors (CPU), memory and bus. The image recognition system may also include input / output interface, network interface and the like.

[0114] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip. The memory is an example of a computer-readable medium.

[0115] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage computer readable storage media or any other non-transmission media that can be used to store information that can be accessed by the image recognition system. As defined in this article, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0116] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or computer-readable storage medium that includes a series of elements includes not only those elements, but also other elements that are not explicitly listed, or also includes elements inherent to such process, method, commodity or computer-readable storage medium. In the absence of further restrictions, the elements defined by the sentence "comprises a ..." do not exclude the presence of other identical elements in the process, method, commodity or computer-readable storage medium that includes the elements.

[0117] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0118] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. An image recognition method applied to chip layout, It is characterized in that Applied to an image recognition system, the method comprises: Acquire a target chip layout, wherein the target chip layout includes a first region layout image and a second region layout image; Performing an empty bridge layout feature mining operation on the target chip layout to obtain a mining result, wherein the empty bridge layout feature mining operation is used to extract empty bridge layout features from a regional layout image binary consisting of the first regional layout image and the second regional layout image, wherein the empty bridge layout feature includes a chip skeleton line block binary consisting of a first chip skeleton line block of the first regional layout image and a second chip skeleton line block of the second regional layout image, and the spatial feature commonality scores of the first chip skeleton line block and the second chip skeleton line block meet a set commonality discrimination condition; On the basis that the mining result indicates that there are empty bridge layout features in the regional map image binary, generating empty bridge layout feature annotation data based on the mining result; The target chip layout is spatially characterized in state by analyzing the spatial feature state according to the air bridge layout feature annotation data to obtain a spatial feature state analysis viewpoint, wherein the spatial feature state analysis viewpoint represents a state conflict weight between the first area layout image and the second area layout image.

2. The method according to claim 1, It is characterized in that The performing of an empty bridge layout feature mining operation on the target chip layout to obtain a mining result includes: Extracting the empty bridge layout features in a preset image data set, wherein the preset image data set includes at least one associated chip skeleton line block binary; If there is a target-associated chip skeleton line map block binary in the preset image data set, a mining result including the target-associated chip skeleton line map block binary is obtained, and the target-associated chip skeleton line map block binary includes the first chip skeleton line map block and the second chip skeleton line map block; wherein each of the associated chip skeleton line map block binary matches the following conditions at the same time: the associated chip skeleton line map block binary includes a main chip main skeleton line map block and a target chip skeleton line map block; the main chip main skeleton line map block and the target chip skeleton line map block correspond to a main area layout image and a target area layout image from an associated area layout image binary, respectively; the associated area layout image binary is identified as a non-associated area layout image in the first spatial feature state parsing network; The main chip main skeleton line map block in the main area layout image is updated to the target chip skeleton line map block to obtain an updated associated area layout image binary group, and the updated associated area layout image binary group is identified as an associated area layout image in the first spatial feature state parsing network.

3. The method according to claim 1 or 2, It is characterized in that The mining result includes at least one chip skeleton line block binary group, and the generating of the empty bridge layout feature annotation data based on the mining result includes: generating a block coding feature vector according to each chip skeleton line block binary group; generating a first state parsing instruction, the first state parsing instruction is used to characterize that the second space feature state parsing network performs a space feature state parsing project according to the block coding feature vector; generating the empty bridge layout feature annotation data based on the block coding feature vector and the first state parsing instruction; The method of performing spatial feature state analysis on the target chip layout based on the empty bridge layout feature annotation data to obtain the spatial feature state analysis viewpoint includes: entering the empty bridge layout feature annotation data and the target chip layout into the second spatial feature state analysis network to obtain the spatial feature state analysis viewpoint.

4. The method according to claim 3, It is characterized in that The method further comprises: On the basis that the mining result indicates that there is no empty bridge layout feature in the target chip layout, a second state parsing instruction is generated, wherein the second state parsing instruction is used to indicate that the second space feature state parsing network directly executes the space feature state parsing project; The second state analysis instruction and the target chip layout are entered into the second spatial feature state analysis network to obtain the spatial feature state analysis viewpoint.

5. The method according to claim 4, It is characterized in that The method includes: the second spatial feature state parsing network performs the following processing: Performing a wavelet vector mining operation on a chip layout set to obtain a first chip layout wavelet vector, wherein the chip layout set is a set obtained by combining target instruction data, the first area layout image, and the second area layout image, wherein the target instruction data includes the empty bridge layout feature labeling data or the second state parsing instruction; Performing a residual wavelet vector mining operation on the first chip layout wavelet vector to obtain a second chip layout wavelet vector; The target subvector of the second chip layout wavelet vector is subjected to spatial feature state mining to obtain the spatial feature state analysis viewpoint, wherein the spatial feature state analysis viewpoint represents the spatial feature commonality score of the first area layout image and the second area layout image.

6. The method according to claim 5, It is characterized in that The second spatial feature state parsing network is debugged based on the following ideas: Acquire a historical chip layout, wherein the historical chip layout includes a first historical region layout image, a second historical region layout image, and training annotations, wherein the training annotations represent state conflict between the first historical region layout image and the second historical region layout image; Performing an empty bridge layout feature mining operation on the historical chip layout to obtain a historical mining result; If the historical mining result indicates that there is no empty bridge layout feature in the historical chip layout, determining the second state parsing instruction as historical instruction data; If the historical mining result indicates that there are empty bridge layout features in the historical chip layout, generating historical empty bridge layout feature annotation data; Determining the historical instruction data according to the historical empty bridge layout feature annotation data; Entering the historical instruction data and the historical chip layout into the second spatial feature state analysis network to obtain a historical spatial feature state analysis viewpoint; Based on the difference between the historical spatial feature state parsing views and the training annotations, the neural network weights of the second spatial feature state parsing network are optimized.

7. The method according to claim 2, It is characterized in that The method further comprises: Acquire multiple positive samples, in which the main historical area map image and the target historical area map image have state conflict; Entering each of the positive examples into the first spatial feature state analysis network to obtain a corresponding spatial feature state analysis viewpoint, and determining the positive examples that do not have state conflict represented by the spatial feature state analysis viewpoint as target positive examples; For each of the target positive examples, the first target chip skeleton line map block in the main historical area layout image in the target positive example is updated to the second target chip skeleton line map block in the target historical area layout image to obtain a reference main historical area layout image; if the spatial feature state analysis viewpoint obtained by entering the reference main historical area layout image and the target historical area layout image into the first spatial feature state analysis network represents that there is a state conflict, the first target chip skeleton line map block and the second target chip skeleton line map block are determined as an associated chip skeleton line map block tuple, and the associated chip skeleton line map block tuple is added to the preset image data set; Among them, the first target chip skeleton line map block is any chip skeleton line map block of the main historical area layout image, and the second target chip skeleton line map block is any chip skeleton line map block of the target historical area layout image.

8. The method according to claim 1, It is characterized in that The method further comprises: The target chip layout is adjusted based on the spatial feature state analysis viewpoint.

9. The method according to claim 1, It is characterized in that The method further comprises: The target chip layout is stored based on the spatial feature state analysis viewpoint.

10. An image recognition system, It is characterized in that It comprises a processor and a memory and a bus connected to the processor; wherein the processor and the memory communicate with each other through the bus; the processor is used to call the program instructions in the memory to execute the image recognition method applied to chip layout as described in any one of claims 1-9.

Citation Information

Patent Citations

  • 2*2 display receiving integrated chip for terahertz radar and detection imaging

    CN114488137A

  • Preparation method of superconductive quantum chip air bridge

    CN114914194A

  • Air bridge three-dimensional modeling method and system for chip layout, medium and equipment

    CN116451635A

  • Method, device and equipment for laying empty bridge of chip layout and storage medium

    CN117852472A

  • Air bridge laying method, device and equipment in circuit layout, medium and product

    CN117993349A