Laser chip heat sink image optimization processing method, system, equipment and medium

Abstract

The invention relates to a laser chip heat sink image optimization processing method, system and device, and a medium. The method comprises the following steps: identifying and extracting a defect subgraph from a heat sink image; reconstructing a pixel value of the defect sub-image to balance gray level distribution, introducing top-hat transform and a filter to optimize and adjust the defect sub-image, and outputting the defect sub-image after optimization processing; carrying out at least one type of image enhancement processing on the defect sub-image after optimization processing; and in response to a fusion instruction of a user, fusing the enhanced defect sub-images with the selected background image so as to realize that a single enhanced defect sub-image is fused into a plurality of regions of the background image or a plurality of enhanced defect sub-images are fused into a single region of the background image. According to the method, the processing effect of the heat sink image of the laser chip is remarkably improved through operations such as accurate identification, optimization adjustment, enhancement processing and flexible fusion, and powerful technical support is provided for links such as defect detection, quality control and process improvement in the laser chip industry.

Description

Technical Field

[0001] The present invention relates to the field of image processing technology, and in particular to a laser chip heat sink image optimization processing method, system, equipment and medium. Background Art

[0002] In the field of chip manufacturing, with the complexity of design and the diversification of processes, the chip output of each production batch is limited, resulting in a significant shortage of sample data available for model training. Directly using these limited data for model training often leads to limited generalization ability of the model and prone to overfitting problems. Although small sample learning methods can alleviate this problem to a certain extent, their effects are still not comparable to training based on large amounts of data. In this case, customers' chips may suffer a lot of losses due to model misjudgment.

[0003] In addition, each process in the chip manufacturing process may introduce large fluctuations, which will directly affect the quality of the final chip. For example, differences in external materials may lead to differences in the color of chip light emission, or small batch errors in the process may produce defective products. These quality fluctuations require that corresponding defective samples be considered in advance and added in model training.

[0004] At the same time, different types of chips have different characteristics and defect modes, which requires training specific detection models for each type of chip. However, due to the limited number of samples of each type of chip, it becomes impractical to directly train a dedicated model, and the importance of data enhancement technology is highlighted at this time.

[0005] In addition, the diversity of inspection equipment and many factors in the imaging process (such as lighting conditions, shooting angles, etc.) will lead to differences in image quality. Such differences make it difficult for the model to cover all possible situations during training, resulting in inconsistent inspection results, increasing the risk of missed inspections and misjudgments, and ultimately affecting the overall product yield.

[0006] Furthermore, the inconsistent roughness of the heat sink and the difficulty in identifying surface defects pose great challenges to model training, which is also an important factor leading to poor model results. Summary of the invention

[0007] 1. Technical issues to be resolved

[0008] In view of the above-mentioned shortcomings and deficiencies of the prior art, the present invention provides a laser chip heat sink image optimization processing method, system, equipment and medium, which solves the technical problems of inconsistent roughness of laser chip heat sinks, difficulty in distinguishing surface defects, complexity of the process, and difficulty in identification caused by a large number of chip types and defect types.

[0009] (II) Technical solution

[0010] In order to achieve the above object, the main technical solutions adopted by the present invention include:

[0011] In a first aspect, an embodiment of the present invention provides a laser chip heat sink image optimization processing method, comprising:

[0012] identifying and extracting defect sub-images from a heat sink image to be processed;

[0013] Reconstruct the pixel values ​​of the defect sub-image to balance the grayscale distribution, introduce top-hat transformation and filter to optimize and adjust the defect sub-image, and output the optimized defect sub-image;

[0014] Performing at least one image enhancement process on the optimized defect sub-image;

[0015] In response to a user's fusion instruction, the enhanced defect sub-image is fused with the selected background image to achieve fusion of a single enhanced defect sub-image into multiple regions of the background image or fusion of multiple enhanced defect sub-images into a single region of the background image.

[0016] Optionally, identifying and extracting defect sub-images from the heat sink image to be processed includes:

[0017] Performing preprocessing including denoising and smoothing filtering on the heat sink image to be processed;

[0018] By using the Canny algorithm, the gradient intensity and direction of each pixel in the heat sink image to be processed after preprocessing are calculated to determine the candidate edge points;

[0019] According to the set high and low thresholds, edges and non-edges including strong edges and weak edges are determined; wherein non-edges refer to edge pixels whose gradient strength is lower than the low threshold, strong edges refer to edge pixels whose gradient strength is higher than the high threshold, and weak edges refer to edge pixels whose gradient strength is between the low threshold and the high threshold and whose eight-neighborhood pixels have a non-edge pixel ratio greater than the set ratio;

[0020] Starting from the strong edge, edge tracking is performed, and the weak edges connected to the strong edge are incorporated into the edge contour to form the final edge point set;

[0021] Based on the final edge point set and combined with the prior knowledge of the acquired heat sink image, the ROI area containing potential defects is determined to generate a defect sub-map.

[0022] Optionally, reconstructing the pixel values ​​of the defect sub-image to balance the grayscale distribution includes:

[0023] Calculate the grayscale histogram of the defect sub-image, and obtain the number of pixels at each grayscale level by traversing each pixel of the defect sub-image and recording the corresponding grayscale value;

[0024] Analyze the distribution characteristics of the grayscale histogram and identify the grayscale peak area and valley area;

[0025] According to the grayscale peak area and valley area and the number of pixels at each grayscale level, an adjustment term is introduced to determine the mapping function;

[0026] The determined mapping function is used to perform pixel-by-pixel distribution processing on the defect sub-image, and the gray value of each pixel is mapped to a new gray value to reconstruct the defect sub-image;

[0027] The mapping function is:

[0028]

[0029] In the formula, S k is the grayscale of the output image, T is the mapping function, r k is the grayscale of the input image, k is the grayscale index, n k is the number of pixels with gray level k, n j is the number of pixels with gray level j, n i is the number of pixels with gray level i, N is the total number of pixels in the image, and L is the number of gray levels.

[0030] Optionally, a top hat transformation and a filter are introduced to optimize and adjust the defect sub-graph, and the defect sub-graph after output optimization processing includes:

[0031] Divide the reconstructed defect sub-image into several local areas, and for each local area, calculate the grayscale feature value including the mean, median and standard deviation;

[0032] Using the calculated grayscale eigenvalues, the size of the structural element corresponding to each local area is dynamically determined;

[0033] According to the structural element corresponding to each local area, an opening operation is performed on the corresponding local area to obtain a background estimation map;

[0034] Subtract the defect sub-image from the background estimation image obtained by the opening operation pixel by pixel to obtain a preliminarily processed defect sub-image;

[0035] The Gabor filter is used to filter the initially processed defect sub-image to respond to the edge information in the image, and the optimized defect sub-image is output;

[0036] Among them, the Gabor filter is:

[0037]

[0038] Wherein, x'=xcosθ+ysinθ, y'=-xsinθ+ycosθ, λ is the wavelength of the filter, θ is the direction of the filter texture, ψ is the phase offset of the filter, γ is the spatial aspect ratio of the filter, σ is the variation rate of the filter, i is the imaginary unit, and x and y represent the horizontal and vertical positions in the image, respectively.

[0039] Optionally, performing at least one image enhancement process on the optimized defect sub-image includes:

[0040] Calculate the mean of the defect subgraph after optimization;

[0041] The corresponding Gamma value is set according to the mean value, wherein when the mean value is greater than a preset threshold value, the Gamma value is set above 1.0, and when the mean value of the defect sub-graph is less than another preset threshold value, the Gamma value is set between 0 and 1.0;

[0042] Gamma correction is performed on the defect sub-image using the set Gamma value to obtain a Gamma-corrected defect sub-image;

[0043] Perform logarithmic transformation on the defect sub-image after Gamma correction to convert the multiplicative relationship of the image into an additive relationship to separate brightness and reflectivity;

[0044] Perform fast Fourier transform on the defect sub-image after logarithmic transformation and enhance the high-frequency components;

[0045] Perform inverse Fourier transform on the enhanced image and perform exponential transformation to obtain the defect sub-image after homomorphic filtering;

[0046] One or more of rotation transformation, scaling transformation or flipping transformation is selected to adjust the spatial position, shape or direction of the defect sub-image after homomorphic filtering processing to obtain the defect sub-image after image enhancement processing.

[0047] Optionally, in response to a fusion instruction from a user, the enhanced defect sub-image is fused with the selected background image to achieve fusion of a single enhanced defect sub-image into multiple regions of the background image or fusion of multiple enhanced defect sub-images into a single region of the background image, including:

[0048] Receive and analyze the fusion instructions input by the user, determine the defect sub-image to be fused, the selected background image, the fusion area and the fusion method;

[0049] According to the fusion instruction, defect information including the shape, size, color and texture of the defect in the selected area is extracted from the defect sub-image;

[0050] If the fusion method is to fuse a single defect sub-image into multiple regions of the background image, the selected defect sub-image is split or copied to ensure that each region to be fused has corresponding defect information;

[0051] If the fusion method is to fuse multiple defect sub-images into a single area, each sub-image is adjusted at least in size and position to ensure that it can be seamlessly spliced ​​into the selected background image;

[0052] According to the fusion instruction or preset rules, one or more areas are marked on the selected background image as the areas to be fused;

[0053] Applying a fusion operation, if the fusion method is to fuse a single defect sub-image into multiple regions of the background image, different parts of the selected defect sub-image are respectively mapped to corresponding regions of the background image; if multiple defect sub-images are fused into a single region of the background image, a preset superposition or combination strategy is adopted to merge the multiple defect sub-images in the selected background image region;

[0054] If repeated fusion is detected, that is, the defect feature map to be fused is similar or identical to the defect feature map that has been fused, the user will be automatically prompted and adjustments or replacements will be made based on the user's feedback;

[0055] Output the fused image to the specified display device or storage medium.

[0056] Optionally, a preset superposition or combination strategy is used to merge multiple defect sub-images within a selected background image region, including:

[0057] Map the color space of the defect sub-image to the color space of the background image;

[0058] Through image registration, multiple defect sub-images are aligned with the background image in space and scale;

[0059] Create an independent layer with editable properties for each aligned defect sub-image;

[0060] Analyze the correlation indicators including position relationship, shape similarity, color and contrast between each defect sub-image, and output the stacking order of the layers based on the preset evaluation criteria and correlation indicators;

[0061] According to the stacking order of the layers, multiple defect sub-images are stacked in the form of different layers on the same background image area;

[0062] Show users a real-time preview of the fusion results, and adjust the overlay order and transparency of each layer in real time based on user feedback or preset conditions.

[0063] In a second aspect, an embodiment of the present invention provides a laser chip heat sink image optimization processing system, comprising:

[0064] The recognition and extraction module is used to recognize and extract defect sub-images from the heat sink image to be processed;

[0065] The optimization processing module is used to reconstruct the pixel values ​​of the defect sub-image to balance the grayscale distribution, introduce the top hat transformation and filter to optimize and adjust the defect sub-image, and output the optimized defect sub-image;

[0066] An image enhancement module, used for performing at least one image enhancement process on the defect sub-image after the optimization process;

[0067] The image fusion module is used to fuse the enhanced defect sub-image with the selected background image in response to the user's fusion instruction, so as to achieve the fusion of a single enhanced defect sub-image into multiple areas of the background image or the fusion of multiple enhanced defect sub-images into a single area of ​​the background image.

[0068] In a third aspect, an embodiment of the present invention provides a laser chip heat sink image optimization processing device, comprising: at least one database; and a memory communicatively connected to the at least one database; wherein the memory stores instructions executable by the at least one database, and the instructions are executed by the at least one database so that the at least one database can execute the laser chip heat sink image optimization processing method as described above.

[0069] In a fourth aspect, an embodiment of the present invention provides a computer-readable medium having computer executable instructions stored thereon, and when the executable instructions are executed by a processor, the laser chip heat sink image optimization processing method as described above is implemented.

[0070] (III) Beneficial effects

[0071] The beneficial effects of the present invention are:

[0072] First, the present invention provides a solid foundation for subsequent image analysis and defect location by accurately identifying and extracting defect sub-images from the heat sink image to be processed. The effectiveness of this step is due to the algorithm's perfect capture of edge features and accurate detection of closed contours, thus ensuring that the extracted defect sub-images are highly representative and authentic.

[0073] Furthermore, the present invention balances the grayscale distribution by reconstructing the pixel values ​​of the defect sub-image. This processing not only enhances the overall contrast of the image, but also significantly improves the ability to express details. In particular, for those areas that were originally darker or obscured in the image, their features are effectively highlighted through this processing, providing more abundant information for subsequent analysis and judgment.

[0074] It is worth mentioning that the present invention also innovatively introduces top-hat transformation and filters to fine-tune the distribution of light and dark areas, contrast, and brightness of the defect sub-image. This optimization process not only further improves the visual effect of the image, but also makes the defect features clearer and more identifiable. This is of great significance for subsequent quality assessment, defect classification, and process improvement.

[0075] In addition, the present invention also provides an image enhancement processing option to further enhance the defect sub-image that has been optimized. This enhancement processing can be flexibly selected according to actual needs, thereby ensuring that the output image can meet the specific requirements of different application scenarios.

[0076] Finally, in response to the user's fusion instruction, the present invention can efficiently fuse the enhanced defect sub-image with the selected background image. This fusion method supports both the fusion of a single enhanced defect sub-image into multiple regions of the background image and the fusion of multiple enhanced defect sub-images into a single region of the background image. This flexibility not only improves the efficiency and convenience of image processing, but also provides users with more diverse display and analysis options.

[0077] Therefore, the present invention significantly improves the processing effect of laser chip heat sink images through a series of innovative steps such as precise identification, optimized adjustment, enhanced processing and flexible fusion, and provides strong technical support for key links such as defect detection, quality control and process improvement in the laser chip industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] Figure 1 A schematic diagram of a process flow of a laser chip heat sink image optimization processing method provided by an embodiment of the present invention;

[0079] Figure 2 A specific flow chart of step S1 of a laser chip heat sink image optimization processing method provided by an embodiment of the present invention;

[0080] Figure 3 A defect sub-graph of a laser chip heat sink image optimization processing method provided by an embodiment of the present invention;

[0081] Figure 4 A specific flow chart of step S2 of a laser chip heat sink image optimization processing method provided by an embodiment of the present invention;

[0082] Figure 5 A reconstructed defect sub-graph of a laser chip heat sink image optimization processing method provided by an embodiment of the present invention;

[0083] Figure 6Another specific flow chart of step S2 of a laser chip heat sink image optimization processing method provided by an embodiment of the present invention;

[0084] Figure 7 A defect sub-image after optimization processing of a laser chip heat sink image optimization processing method provided by an embodiment of the present invention;

[0085] Figure 8 A specific flow chart of step S3 of a laser chip heat sink image optimization processing method provided by an embodiment of the present invention;

[0086] Fig. 9 A defect sub-image after Gamma correction of a laser chip heat sink image optimization processing method provided by an embodiment of the present invention;

[0087] Fig.10 A defect sub-image after homomorphic filtering processing of a laser chip heat sink image optimization processing method provided by an embodiment of the present invention;

[0088] Fig.11 A defect sub-image after image enhancement processing of a laser chip heat sink image optimization processing method provided by an embodiment of the present invention;

[0089] Fig.12 A specific flow chart of step S4 of a laser chip heat sink image optimization processing method provided by an embodiment of the present invention;

[0090] Fig.13 A specific flow chart of step S45 of a laser chip heat sink image optimization processing method provided by an embodiment of the present invention;

[0091] Fig.14 A fused image of a laser chip heat sink image optimization processing method provided by an embodiment of the present invention;

[0092] Fig.15 A schematic diagram of image enhancement processing of different defect types in a laser chip heat sink image optimization processing method provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0093] In order to better explain the present invention and facilitate understanding, the present invention is described in detail below through specific implementation modes in conjunction with the accompanying drawings.

[0094] like Figure 1As shown, an optimization processing method for a laser chip heat sink image proposed in an embodiment of the present invention includes: identifying and extracting defect sub-images from a heat sink image to be processed; reconstructing pixel values ​​of the defect sub-images to balance the grayscale distribution, and introducing a top-hat transform and a filter to optimize and adjust the defect sub-images, and outputting the optimized defect sub-images; performing at least one image enhancement processing on the optimized defect sub-images; and in response to a user's fusion instruction, fusing the enhanced defect sub-images with a selected background image to achieve fusion of a single enhanced defect sub-image into multiple regions of the background image or fusion of multiple enhanced defect sub-images into a single region of the background image.

[0095] First, the present invention provides a solid foundation for subsequent image analysis and defect location by accurately identifying and extracting defect sub-images from the heat sink image to be processed. The effectiveness of this step is due to the algorithm's perfect capture of edge features and accurate detection of closed contours, thus ensuring that the extracted defect sub-images are highly representative and authentic.

[0096] Furthermore, the present invention balances the grayscale distribution by reconstructing the pixel values ​​of the defect sub-image. This processing not only enhances the overall contrast of the image, but also significantly improves the ability to express details. In particular, for those areas that were originally darker or obscured in the image, their features are effectively highlighted through this processing, providing more abundant information for subsequent analysis and judgment.

[0097] It is worth mentioning that the present invention also innovatively introduces top-hat transformation and filters to fine-tune the distribution of light and dark areas, contrast, and brightness of the defect sub-image. This optimization process not only further improves the visual effect of the image, but also makes the defect features clearer and more identifiable. This is of great significance for subsequent quality assessment, defect classification, and process improvement.

[0098] In addition, the present invention also provides an image enhancement processing option to further enhance the defect sub-image that has been optimized. This enhancement processing can be flexibly selected according to actual needs, thereby ensuring that the output image can meet the specific requirements of different application scenarios.

[0099] Finally, in response to the user's fusion instruction, the present invention can efficiently fuse the enhanced defect sub-image with the selected background image. This fusion method supports both the fusion of a single enhanced defect sub-image into multiple regions of the background image and the fusion of multiple enhanced defect sub-images into a single region of the background image. This flexibility not only improves the efficiency and convenience of image processing, but also provides users with more diverse display and analysis options.

[0100] Therefore, the present invention significantly improves the processing effect of laser chip heat sink images through a series of innovative steps such as precise identification, optimized adjustment, enhanced processing and flexible fusion, and provides strong technical support for key links such as defect detection, quality control and process improvement in the laser chip industry.

[0101] In order to better understand the above technical solution, exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided to enable a clearer and more thorough understanding of the present invention and to fully convey the scope of the present invention to those skilled in the art.

[0102] Specifically, an embodiment of the present invention provides a laser chip heat sink image optimization processing method, comprising:

[0103] S1. Identify and extract defect sub-images from the heat sink image to be processed.

[0104] Furthermore, if Figure 2 As shown, step S1 includes:

[0105] S11, performing preprocessing including denoising and smoothing filtering on the heat sink image to be processed.

[0106] S12. Calculate the gradient intensity and direction of each pixel in the heat sink image to be processed after preprocessing by using the Canny algorithm to determine candidate edge points.

[0107] S13. According to the set high and low thresholds, determine the edges and non-edges including strong edges and weak edges; wherein the non-edge refers to the edge pixel point whose gradient strength is lower than the low threshold, the strong edge refers to the edge pixel point whose gradient strength is higher than the high threshold, and the weak edge refers to the edge pixel point whose gradient strength is between the low threshold and the high threshold and whose eight-neighborhood pixel ratio and non-edge pixel ratio are greater than the set ratio.

[0108] S14, starting from the strong edge, edge tracking is performed, and weak edges connected to the strong edge are incorporated into the edge contour, thereby forming a final edge point set.

[0109] S15. Based on the final edge point set and in combination with the prior knowledge of the acquired heat sink image, determine the ROI region containing potential defects to generate a defect sub-image.

[0110] However, reference Figure 3It can be observed that the obtained defect sub-image will lose some edge information features in some cases, which may lead to false defect judgments. This problem is particularly obvious in environments with uneven lighting. Therefore, it is necessary to further process the edge features to improve the accuracy of defect detection.

[0111] S2. Reconstruct the pixel values ​​of the defect sub-image to balance the grayscale distribution, introduce top-hat transformation and filter to optimize and adjust the defect sub-image, and output the optimized defect sub-image.

[0112] Furthermore, if Figure 4 As shown, the pixel values ​​of the reconstructed defect sub-image in step S2 to balance the grayscale distribution include:

[0113] S21, calculating the grayscale histogram of the defect sub-image, and obtaining the number of pixels at each grayscale level by traversing each pixel of the defect sub-image and recording the corresponding grayscale value.

[0114] First, a grayscale histogram is initialized, the horizontal axis of which represents different grayscale levels (for example, in an 8-bit grayscale image, the grayscale level ranges from 0 to 255), and the vertical axis is used to record the number of pixels that appear at the corresponding grayscale level.

[0115] Next, traverse each pixel in the defect sub-image. For each pixel, read its grayscale value and find the corresponding grayscale level in the grayscale histogram. Then, increase the pixel count of that grayscale level by 1 to indicate that a pixel with that specific grayscale value has been encountered.

[0116] By traversing the entire defect sub-image and repeating the above process, a complete grayscale histogram is finally obtained, in which each grayscale has a corresponding number of pixels. This histogram provides important information about the grayscale distribution in the image and is the basis for subsequent image analysis.

[0117] S22. Analyze the distribution characteristics of the grayscale histogram and identify the grayscale peak area and valley area.

[0118] After obtaining the grayscale histogram, its distribution characteristics need to be further analyzed, which includes identifying the peak area and valley area in the grayscale histogram.

[0119] Peak areas refer to grayscale areas with relatively high number of pixels in the grayscale histogram. These areas usually correspond to the main color or brightness areas in the image, and may represent the background, objects, or other important features in the image. By analyzing the peak areas, we can understand the main grayscale components in the image and their distribution.

[0120] Valley areas refer to grayscale areas with relatively low number of pixels in the grayscale histogram. These areas may be located between two peaks, representing the transition or boundary between different grayscale areas in the image. Analysis of valley areas helps to identify details and edge information in the image, especially in defect detection, where valley areas may directly correspond to potential defects or abnormal areas.

[0121] S23, according to the grayscale peak area and valley area and the number of pixels at each grayscale level, and introducing an adjustment item, a mapping function is determined. The mapping function is:

[0122]

[0123] In the formula, S k is the grayscale of the output image, T is the mapping function, r k is the grayscale of the input image, k is the grayscale index, n k is the number of pixels with gray level k, n j is the number of pixels with gray level j, n i is the number of pixels with gray level i, N is the total number of pixels in the image, and L is the number of gray levels.

[0124] The purpose of this adjustment is to reduce the over-stretching effect that may occur during the histogram equalization process, thereby better preserving the local details of the image.

[0125] Specifically, histogram equalization is a commonly used image enhancement technique that can enhance the overall contrast of an image by adjusting the grayscale distribution of the image. However, in some cases, histogram equalization may cause some local areas of the image to be over-stretched, thereby losing some important detail information.

[0126] To solve this problem, the mapping function introduces this adjustment term. This adjustment term takes into account the difference between the number of pixels at each gray level in the image and the ideal uniform distribution, and adjusts the mapping value of each gray level according to this difference. In this way, the mapping function can better preserve the local details of the image while enhancing the contrast of the image.

[0127] Therefore, this adjustment term plays a balancing role in the formula, which dynamically adjusts the behavior of the mapping function according to the gray level distribution of the image, thereby achieving better image enhancement effects.

[0128] S24, using the determined mapping function to perform pixel-by-pixel allocation processing on the defect sub-image, mapping the grayscale value of each pixel to a new grayscale value, so as to reconstruct the defect sub-image.

[0129] like Figure 5As shown in the figure, by redistributing the pixel values ​​to reconstruct the defect sub-image, a more uniform distribution of the image grayscale in each interval can be achieved. Such processing results not only improve the contrast of the image edge, making the edge features more obvious and easier to identify, but also improve the overall contrast and detail expression of the image. This means that the reconstructed defect sub-image will provide more accurate and richer information in subsequent defect detection and analysis.

[0130] Then, if Figure 6 As shown, in step S2, top hat transformation and filter are introduced to optimize and adjust the defect sub-graph, and the defect sub-graph after output optimization processing includes:

[0131] S25, dividing the reconstructed defect sub-image into a number of local areas, and for each local area, calculating a grayscale feature value including a mean value, a median, and a standard deviation.

[0132] S26. Using the calculated grayscale eigenvalues, dynamically determine the size of the structural element corresponding to each local area.

[0133] First, the reconstructed defect sub-image is subdivided into multiple local regions. Such division helps to more accurately analyze the grayscale features of each region and adjust the parameters of subsequent processing accordingly.

[0134] For each divided local area, calculate its grayscale feature values, including the mean, median and standard deviation. The mean reflects the overall level of grayscale in the area, the median represents the central trend of the grayscale, and the standard deviation reveals the discrete degree or fluctuation range of the grayscale value.

[0135] The size of the structural element corresponding to each local area is dynamically determined using these calculated grayscale feature values. The structural element is the basic unit in morphological operations, and the selection of its size directly affects the effect of morphological processing such as opening operations. By dynamically adjusting the size of the structural element according to the grayscale characteristics of the local area, it can more flexibly adapt to the characteristics of different areas in the image, thereby improving the accuracy and effectiveness of subsequent processing.

[0136] S27. Perform an opening operation on the corresponding local area according to the structural element corresponding to each local area, so as to obtain a background estimation map.

[0137] S28, subtracting the defect sub-image from the background estimation image obtained through the opening operation pixel by pixel to obtain a preliminarily processed defect sub-image.

[0138] According to the structural element corresponding to each local area determined in step S26, an opening operation is performed on the corresponding local area. The opening operation is a basic operation in morphological processing, which eliminates small bright spots in the image by a process of first dilation and then corrosion, while keeping the overall grayscale and larger bright areas unchanged. The image processed in this way can be used as a background estimation map, which reflects the background information after removing the small bright spots in the original image.

[0139] Next, the original defect sub-image is subtracted pixel by pixel from the background estimation image obtained by the opening operation. The morphological transformation of the dark area or details in the image is highlighted by the difference between the original image and the opening operation result. In this way, the dark features in the defect sub-image can be further extracted, including details such as holes and cracks, so as to obtain a clearer and more accurate representation of the defect sub-image.

[0140] S29, using a Gabor filter to filter the initially processed defect sub-image to respond to edge information in the image, and outputting an optimized defect sub-image.

[0141] The parameters of the Gabor filter are set according to the characteristics of the defect sub-image and the type of edge information to be extracted. These parameters include wavelength, direction, phase shift, standard deviation of the Gaussian function, and spatial aspect ratio. The choice of parameters is determined experimentally to achieve the best filtering effect. For example, the wavelength and direction parameters can be adjusted to match the specific frequency and direction of the edge in the defect sub-image.

[0142] The defect sub-image processed initially is used as input and filtered by the set Gabor filter. The Gabor filter can respond to the edge information in the image and suppress the non-edge information by locating in both the frequency domain and the spatial domain.

[0143] Among them, the Gabor filter is:

[0144]

[0145] Wherein, x'=xcosθ+ysinθ, y'=-xsinθ+ycosθ, λ is the wavelength of the filter, θ is the direction of the filter texture, ψ is the phase offset of the filter, γ is the spatial aspect ratio of the filter, σ is the variation rate of the filter, i is the imaginary unit, and x and y represent the horizontal and vertical positions in the image, respectively.

[0146] like Figure 7As shown in the figure, after this step of optimization processing, a defect sub-image with significantly improved edge effects is obtained. It can be clearly seen in the figure that the defect edge, which may have been blurred due to uneven lighting or background interference, is now accurately outlined and the details are greatly enhanced. Such processing results undoubtedly provide more accurate and reliable visual support for subsequent defect location, classification and quantitative analysis.

[0147] S3. Perform at least one image enhancement process on the optimized defect sub-image.

[0148] Furthermore, if Figure 8 As shown, step S3 includes:

[0149] S31, calculating the mean of the defect sub-image after optimization. This mean reflects the overall brightness level of the image. This mean can be obtained by traversing all pixels of the defect sub-image and accumulating their values, and then dividing by the total number of pixels.

[0150] S32. Set a corresponding Gamma value according to the mean, wherein when the mean is greater than a preset threshold, the Gamma value is set to be greater than 1.0, and when the mean of the defect sub-graph is less than another preset threshold, the Gamma value is set to be between 0 and 1.0.

[0151] In step S32, the Gamma value is dynamically set according to the mean value calculated in the previous step. The Gamma value is a key parameter that determines the adjustment range of image brightness and contrast. The formula used in the present invention is Ioutput = I γ input The gamma enhancement formula is to enhance or reduce the brightness and contrast by adjusting or inputting the power of the pixel value of the image. Among them, γ is the key parameter to control the brightness and contrast of the image. By adjusting the Gamma value, the contrast of the image can be enhanced.

[0152] When the mean value is greater than a preset high threshold (e.g. 150), it means that the image is bright as a whole. In this case, the Gamma value is set above 1.0 to enhance the details and contrast of the bright parts of the image. When the mean value is less than another preset low threshold (e.g. 100), it means that the image is dark as a whole. In this case, the Gamma value is set between 0 and 1.0 to enhance the details and contrast of the dark parts of the image. By dynamically adjusting the Gamma value, it is possible to more flexibly adapt to defect sub-images with different brightness levels, thereby achieving more accurate brightness and contrast optimization.

[0153] S33, performing gamma correction on the defective sub-image by using the set gamma value to obtain a gamma-corrected defective sub-image.

[0154] In step S33, the Gamma value set in the previous step is used to perform Gamma correction on the defective sub-image. Specifically, each pixel value in the defective sub-image is raised to a power according to the formula (that is, the new value of each pixel value is equal to the Gamma power of its original value multiplied by a constant, which is usually used to ensure that the brightness range of the corrected image remains unchanged). Through this process, the brightness value of each pixel can be accurately adjusted according to the overall brightness level of the image, thereby achieving global brightness and contrast optimization. By randomly using three groups of different gamma values ​​for correction, three groups of Gamma-corrected defective sub-images are obtained after completion, as shown in FIG. Fig. 9 As shown, it can be clearly seen that the contrast and details of the image have been significantly improved.

[0155] S34, performing logarithmic transformation on the defective sub-image after Gamma correction, converting the multiplicative relationship of the image into an additive relationship, so as to separate the brightness and reflectivity.

[0156] S35, performing fast Fourier transform on the defect sub-image after logarithmic transformation, and enhancing the high-frequency components.

[0157] S36, performing inverse Fourier transform on the enhanced image and performing exponential transform to obtain a defect sub-image after homomorphic filtering processing.

[0158] In a specific embodiment, the image f(x,y)=i(x,y)×r(x,y) is first logarithmically transformed to convert the multiplicative relationship of the image into an additive relationship, thereby obtaining ln(f(x,y))=ln(i(x,y))+ln(r(x,y)). The purpose is to process the brightness and reflectivity of the image separately, because the brightness changes slowly and can be regarded as a low-frequency component, while the reflectivity changes quickly and can be regarded as a high-frequency component; secondly, the logarithmic transformed image is subjected to a fast Fourier transform (FFT) to obtain F(u,v)=I(u,v)+R(u,v), and according to the homomorphic filtering function H(u,v), the high-frequency component is enhanced according to the formula S(u,v)=H(u,v)×F(u,v) in order to compress the dynamic range of the image, enhance its contrast information, and sharpen the edges and details of the image. The processed frequency domain image is converted back to the spatial domain through inverse Fourier transform and exponential transform, and the defect sub-image after homomorphic filtering is obtained.

[0159] Therefore, homomorphic filtering is mainly used for image enhancement, shadow removal and denoising. In terms of image enhancement, it can significantly improve the contrast and detail of the image, making the image clearer. At the same time, by eliminating the influence of shadows, homomorphic filtering can also make the target object easier to identify. Its processing effect is as follows Fig.10 shown.

[0160] S37, selecting one or more of rotation transformation, scaling transformation or flipping transformation to adjust the spatial position, shape or direction of the defect sub-image after homomorphic filtering processing, so as to obtain the defect sub-image after image enhancement processing.

[0161] The present invention uses three enhancement methods, namely rotation transformation, scaling transformation, and flipping transformation. The rotation transformation method is to rotate the figure around a fixed point or axis to change the direction of the figure; scaling transformation is to change the size of the figure, which can be enlarged or reduced; flipping transformation is to flip the figure symmetrically along a certain axis, including horizontal flipping and vertical flipping. Fig.11 As shown, these three methods aim to change the spatial position, shape or orientation of the image to make the data more extensive.

[0162] S4. In response to a fusion instruction from the user, the enhanced defect sub-image is fused with the selected background image to achieve fusion of a single enhanced defect sub-image into multiple regions of the background image or fusion of multiple enhanced defect sub-images into a single region of the background image.

[0163] Furthermore, if Fig.12 As shown, step S4 includes:

[0164] S41, receiving and parsing the fusion instruction input by the user, determining the defect sub-images to be fused, the selected background image, the fusion area and the fusion method. The instruction specifies which defect sub-images need to be fused, which background image is selected as the base, in which specific areas the fusion should occur, and the specific fusion method.

[0165] S42. According to the fusion instruction, defect information including the shape, size, color and texture of the defect in the selected area is extracted from the defect sub-image.

[0166] S43. If the fusion method is to fuse a single defect sub-image into multiple areas of the background image, the selected defect sub-image is split or copied to ensure that each area to be fused has corresponding defect information; if the fusion method is to fuse multiple defect sub-images into a single area, each sub-image is adjusted in at least size and position to ensure that it can be seamlessly spliced ​​into the selected background image.

[0167] S44. According to the fusion instruction or preset rules, one or more regions are marked on the selected background image as regions to be fused.

[0168] S45. Apply a fusion operation. If the fusion method is to fuse a single defect sub-image into multiple areas of the background image, different parts of the selected defect sub-image are respectively mapped to corresponding areas of the background image; if multiple defect sub-images are fused into a single area of ​​the background image, a preset overlay or combination strategy is used to merge the multiple defect sub-images within the selected background image area.

[0169] There are multiple areas in the heat sink image, and each area has different defects in the process. Therefore, you can customize the selected area in the image fusion; you can also fuse multiple defects, and do a regional screening here to prevent defects from being repeatedly fused during fusion. That is, you can set multiple areas for a single defect and a single area for multiple defects.

[0170] Furthermore, if Fig.13 In step S45, the preset superposition or combination strategy is adopted to merge the multiple defect sub-images in the selected background image area, including:

[0171] S451, mapping the color space of the defect sub-image to the color space of the background image.

[0172] S452, aligning multiple defect sub-images with the background image in space and scale through image registration.

[0173] S453. Create an independent layer with editable properties for each aligned defect sub-image. The advantage of this is that the display state and properties of each defect sub-image, such as transparency, size, etc., can be flexibly adjusted in subsequent operations, thereby achieving more refined fusion control.

[0174] S454, analyzing the correlation indicators including position relationship, shape similarity, color and contrast between each defect sub-image, and outputting the stacking order of the layers based on the preset evaluation criteria and correlation indicators. This order determines which defect sub-image should be stacked first and which should be stacked later during the fusion process, thereby ensuring that the final fusion effect meets the requirements of visual perception and logical order.

[0175] S455, multiple defect sub-images are superimposed on the same background image area in different layers according to the layer superposition order. In this process, each layer can adjust properties such as transparency as needed to achieve a better fusion effect.

[0176] S456. Display the real-time preview of the fusion result to the user, and adjust the stacking order and transparency of each layer in real time according to the user's feedback or preset conditions until the fusion effect satisfactory to the user is achieved. This interactive adjustment method not only improves the flexibility of fusion, but also greatly improves the user experience.

[0177] S46. If a repeated fusion is detected, that is, the defect feature map to be fused is similar or identical to the defect feature map that has been fused, the user is automatically prompted and adjustments or replacements are made based on the user's feedback.

[0178] S47: Output the fused image to a designated display device or storage medium.

[0179] This step is to ensure that users can easily view and use the fused image. Fig.14 As shown in FIG. 1 , the image effect after the fusion processing of the above steps is demonstrated, that is, a better edge effect is obtained, and finally a true defect image in which the feature map and the background map are combined into one is achieved.

[0180] Furthermore, users can choose to display the fused image directly on the screen for real-time viewing, or save it to a hard disk, USB flash drive or other storage media for subsequent analysis, processing and sharing, based on actual needs. In this way, the fused image can be effectively presented and utilized, meeting the diverse needs of users in image processing and analysis.

[0181] In addition, in order to more comprehensively demonstrate the strength and effect of image fusion technology, such as Fig.15 As shown in the figure, different defect types (such as splashes in the first row, stains in the second row, and pits in the third row) are used for image enhancement. These enhanced renderings clearly show the detailed features of various defects, further proving the significant role of fusion technology in improving image quality and information recognition.

[0182] In addition, an embodiment of the present invention provides a laser chip heat sink image optimization processing system, including:

[0183] The recognition and extraction module is used to recognize and extract defect sub-images from the heat sink image to be processed.

[0184] The optimization processing module is used to reconstruct the pixel values ​​of the defect sub-image to balance the grayscale distribution, introduce the top hat transformation and filter to optimize and adjust the defect sub-image, and output the optimized defect sub-image.

[0185] The image enhancement module is used to perform at least one image enhancement process on the defect sub-image after optimization.

[0186] The image fusion module is used to fuse the enhanced defect sub-image with the selected background image in response to the user's fusion instruction, so as to achieve the fusion of a single enhanced defect sub-image into multiple areas of the background image or the fusion of multiple enhanced defect sub-images into a single area of ​​the background image.

[0187] Furthermore, an embodiment of the present invention also provides a laser chip heat sink image optimization processing device, comprising: at least one database; and a memory communicatively connected to the at least one database; wherein the memory stores instructions that can be executed by the at least one database, and the instructions are executed by the at least one database so that the at least one database can execute the laser chip heat sink image optimization processing method as described above.

[0188] The laser chip heat sink image optimization processing device shown in the embodiment of the present invention is an advanced system integrating data storage and processing functions. The core components of the device include at least one database and a memory connected to the database.

[0189] The database here is a system for storing, retrieving and managing data. In this device, the database is not only responsible for saving the relevant data of the laser chip heat sink image, but also undertakes the task of executing specific instructions. These instructions are necessary for optimizing the processing of the laser chip heat sink image.

[0190] The memory is a key component closely connected to the database, which stores a series of instructions that can be executed by the database. These instructions are carefully designed and programmed to guide the database on how to efficiently perform the optimization processing of laser chip heat sink images. When the database executes these instructions, it can process the laser chip heat sink images with high quality according to the predetermined methods and processes.

[0191] This laser chip heat sink image optimization processing device achieves efficient and accurate optimization processing of laser chip heat sink images through the coordinated work of its internal database and memory. This processing method can not only improve image quality, but also help improve the performance and stability of laser chips.

[0192] Furthermore, an embodiment of the present invention further provides a computer-readable medium having computer executable instructions stored thereon, and when the executable instructions are executed by a processor, the above-mentioned laser chip heat sink image optimization processing method is implemented.

[0193] The embodiment of the present invention further provides a computer-readable medium, which is a special storage medium that can store various types of digital information. On this computer-readable medium, specific computer-executable instructions are stored. These instructions are pre-written program codes specifically used to implement specific functions.

[0194] Specifically, when these executable instructions stored on computer-readable media are read and executed by the processor, they will guide the processor to optimize the laser chip heat sink image according to predetermined steps and logic. This processing method may include image preprocessing, feature extraction, noise removal, contrast enhancement and other links, aiming to improve the quality and clarity of the laser chip heat sink image, thereby providing a more accurate and reliable data basis for subsequent analysis and application.

[0195] Therefore, through this computer-readable medium and the executable instructions stored thereon, the embodiment of the present invention provides a convenient, efficient and reliable way to achieve the optimization processing of the laser chip heat sink image. This not only helps to improve the application effect of related technologies, but also provides strong technical support for the research and development and production of laser chips.

[0196] Since the system / device described in the above embodiments of the present invention is a system / device used to implement the method of the above embodiments of the present invention, a person skilled in the art can understand the specific structure and deformation of the system / device based on the method described in the above embodiments of the present invention, and thus will not be described in detail here. All systems / devices used in the method of the above embodiments of the present invention belong to the scope of protection of the present invention.

[0197] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention 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 code.

[0198] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems) and computer program products according to embodiments of the present invention. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions.

[0199] It should be noted that in the claims, any reference numerals placed between brackets shall not be construed as limiting the claims. The word "comprising" does not exclude the presence of components or steps not listed in the claims. The word "a" or "an" preceding a component does not exclude the presence of a plurality of such components. The invention may be implemented by means of hardware comprising several different components and by means of a suitably programmed computer. In the claims enumerating several means, several of these means may be embodied by the same hardware. The use of the words first, second, third, etc., is for convenience of expression only and does not indicate any order. These words may be understood as part of the component name.

[0200] In addition, it should be noted that, in the description of this specification, the description of the terms "one embodiment", "some embodiments", "embodiment", "example", "specific example" or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, unless they are contradictory.

[0201] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments after knowing the basic creative concept. Therefore, the claims should be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the present invention.

[0202] Obviously, those skilled in the art can make various modifications and variations to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention should also include these modifications and variations.

Claims

1. A laser chip heat sink image optimization processing method, characterized in that: include: identifying and extracting defect sub-images from a heat sink image to be processed; Reconstruct the pixel values ​​of the defect sub-image to balance the grayscale distribution, introduce top-hat transformation and filter to optimize and adjust the defect sub-image, and output the optimized defect sub-image; Performing at least one image enhancement process on the optimized defect sub-image; In response to a user's fusion instruction, the enhanced defect sub-image is fused with the selected background image to achieve fusion of a single enhanced defect sub-image into multiple regions of the background image or fusion of multiple enhanced defect sub-images into a single region of the background image.

2. The laser chip heat sink image optimization processing method according to claim 1, characterized in that: Identifying and extracting defect sub-images from the heat sink image to be processed includes: Performing preprocessing including denoising and smoothing filtering on the heat sink image to be processed; By using the Canny algorithm, the gradient intensity and direction of each pixel in the heat sink image to be processed after preprocessing are calculated to determine the candidate edge points; According to the set high and low thresholds, edges and non-edges including strong edges and weak edges are determined; wherein non-edges refer to edge pixels whose gradient strength is lower than the low threshold, strong edges refer to edge pixels whose gradient strength is higher than the high threshold, and weak edges refer to edge pixels whose gradient strength is between the low threshold and the high threshold and whose eight-neighborhood pixels have a non-edge pixel ratio greater than the set ratio; Starting from the strong edge, edge tracking is performed, and the weak edges connected to the strong edge are incorporated into the edge contour to form the final edge point set; Based on the final edge point set and combined with the prior knowledge of the acquired heat sink image, the ROI area containing potential defects is determined to generate a defect sub-map.

3. The laser chip heat sink image optimization processing method according to claim 1, characterized in that: Reconstructing the pixel values ​​of the defect sub-image to balance the grayscale distribution includes: Calculate the grayscale histogram of the defect sub-image, and obtain the number of pixels at each grayscale level by traversing each pixel of the defect sub-image and recording the corresponding grayscale value; Analyze the distribution characteristics of the grayscale histogram and identify the grayscale peak area and valley area; According to the grayscale peak area and valley area and the number of pixels at each grayscale level, an adjustment term is introduced to determine the mapping function; The determined mapping function is used to perform pixel-by-pixel distribution processing on the defect sub-image, and the gray value of each pixel is mapped to a new gray value to reconstruct the defect sub-image; The mapping function is: In the formula, S k is the grayscale of the output image, T is the mapping function, r k is the grayscale of the input image, k is the grayscale index, n k is the number of pixels with gray level k, n j is the number of pixels with gray level j, n i is the number of pixels with gray level i, N is the total number of pixels in the image, and L is the number of gray levels.

4. The laser chip heat sink image optimization processing method according to claim 1, characterized in that: The top hat transformation and filter are introduced to optimize and adjust the defect sub-graph. The output optimized defect sub-graph includes: Divide the reconstructed defect sub-image into several local areas, and for each local area, calculate the grayscale feature value including the mean, median and standard deviation; Using the calculated grayscale eigenvalues, the size of the structural element corresponding to each local area is dynamically determined; According to the structural element corresponding to each local area, an opening operation is performed on the corresponding local area to obtain a background estimation map; Subtract the defect sub-image from the background estimation image obtained by the opening operation pixel by pixel to obtain a preliminarily processed defect sub-image; The Gabor filter is used to filter the initially processed defect sub-image to respond to the edge information in the image, and the optimized defect sub-image is output; Among them, the Gabor filter is: Wherein, x'=xcosθ+ysinθ, y'=-xsinθ+ycosθ, λ is the wavelength of the filter, θ is the direction of the filter texture, ψ is the phase offset of the filter, γ is the spatial aspect ratio of the filter, σ is the variation rate of the filter, i is the imaginary unit, and x and y represent the horizontal and vertical positions in the image, respectively.

5. The laser chip heat sink image optimization processing method according to claim 1, characterized in that: Performing at least one image enhancement process on the optimized defect sub-image includes: Calculate the mean of the defect subgraph after optimization; The corresponding Gamma value is set according to the mean value, wherein when the mean value is greater than a preset threshold value, the Gamma value is set above 1.0, and when the mean value of the defect sub-graph is less than another preset threshold value, the Gamma value is set between 0 and 1.0; Gamma correction is performed on the defect sub-image using the set Gamma value to obtain a Gamma-corrected defect sub-image; Perform logarithmic transformation on the defect sub-image after Gamma correction to convert the multiplicative relationship of the image into an additive relationship to separate brightness and reflectivity; Perform fast Fourier transform on the defect sub-image after logarithmic transformation and enhance the high-frequency components; Perform inverse Fourier transform on the enhanced image and perform exponential transformation to obtain the defect sub-image after homomorphic filtering; One or more of rotation transformation, scaling transformation or flipping transformation is selected to adjust the spatial position, shape or direction of the defect sub-image after homomorphic filtering processing to obtain the defect sub-image after image enhancement processing.

6. The laser chip heat sink image optimization processing method according to any one of claims 1 to 5, characterized in that: In response to a fusion instruction from a user, the enhanced defect sub-image is fused with the selected background image to achieve fusion of a single enhanced defect sub-image into multiple regions of the background image or fusion of multiple enhanced defect sub-images into a single region of the background image, including: Receive and analyze the fusion instructions input by the user, determine the defect sub-image to be fused, the selected background image, the fusion area and the fusion method; According to the fusion instruction, defect information including the shape, size, color and texture of the defect in the selected area is extracted from the defect sub-image; If the fusion method is to fuse a single defect sub-image into multiple regions of the background image, the selected defect sub-image is split or copied to ensure that each region to be fused has corresponding defect information; If the fusion method is to fuse multiple defect sub-images into a single area, each sub-image is adjusted at least in size and position to ensure that it can be seamlessly spliced ​​into the selected background image; According to the fusion instruction or preset rules, one or more areas are marked on the selected background image as the areas to be fused; Applying a fusion operation, if the fusion method is to fuse a single defect sub-image into multiple regions of the background image, different parts of the selected defect sub-image are respectively mapped to corresponding regions of the background image; if multiple defect sub-images are fused into a single region of the background image, a preset superposition or combination strategy is adopted to merge the multiple defect sub-images in the selected background image region; If repeated fusion is detected, that is, the defect feature map to be fused is similar or identical to the defect feature map that has been fused, the user will be automatically prompted and adjustments or replacements will be made based on the user's feedback; Output the fused image to the specified display device or storage medium.

7. The laser chip heat sink image optimization processing method according to claim 6, characterized in that: Using a preset superposition or combination strategy, multiple defect sub-images are merged in the selected background image area, including: Map the color space of the defect sub-image to the color space of the background image; Through image registration, multiple defect sub-images are aligned with the background image in space and scale; Create an independent layer with editable properties for each aligned defect sub-image; Analyze the correlation indicators including position relationship, shape similarity, color and contrast between each defect sub-image, and output the stacking order of the layers based on the preset evaluation criteria and correlation indicators; According to the stacking order of the layers, multiple defect sub-images are stacked in the form of different layers on the same background image area; Show users a real-time preview of the fusion results, and adjust the overlay order and transparency of each layer in real time based on user feedback or preset conditions.

8. A laser chip heat sink image optimization processing system, characterized in that: include: The recognition and extraction module is used to recognize and extract defect sub-images from the heat sink image to be processed; The optimization processing module is used to reconstruct the pixel values ​​of the defect sub-image to balance the grayscale distribution, introduce the top hat transformation and filter to optimize and adjust the defect sub-image, and output the optimized defect sub-image; An image enhancement module, used for performing at least one image enhancement process on the defect sub-image after the optimization process; The image fusion module is used to fuse the enhanced defect sub-image with the selected background image in response to the user's fusion instruction, so as to achieve the fusion of a single enhanced defect sub-image into multiple areas of the background image or the fusion of multiple enhanced defect sub-images into a single area of ​​the background image.

9. A laser chip heat sink image optimization processing device, characterized in that: include: at least one database; and a memory in communication with the at least one database; The memory stores instructions that can be executed by the at least one database, and the instructions are executed by the at least one database so that the at least one database can execute the laser chip heat sink image optimization processing method according to any one of claims 1 to 7.

10. A computer-readable medium having computer-executable instructions stored thereon, characterized in that: When the executable instructions are executed by the processor, the laser chip heat sink image optimization processing method as described in any one of claims 1 to 7 is implemented.

Images