Real-time image enhancement method and system based on CD-SEM raster scanning partitioning
The real-time image enhancement method based on CD-SEM raster scanning blocks solves the problems of low contrast and latency in high-resolution images, achieving efficient and low-memory image enhancement that meets the real-time and accuracy requirements of semiconductor detection.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- 48TH RES INST OF CHINA ELECTRONICS TECH GROUP CORP
- Filing Date
- 2026-01-27
- Publication Date
- 2026-05-08
AI Technical Summary
In semiconductor manufacturing, existing technologies such as CD-SEM for high-resolution images suffer from low overall contrast and blurred details, resulting in low accuracy in automatic measurement and defect identification. Furthermore, traditional image enhancement algorithms require huge amounts of memory and processing latency, which cannot meet the needs of real-time detection.
A real-time image enhancement method based on CD-SEM raster scanning blocks is adopted. By presetting the image processing block size and overlapping region parameters, the electron beam scanning is controlled and local histogram statistics, cropping and equalization are performed in real time. Combined with a weighted average algorithm, image fusion is performed to generate an efficient and low-memory enhanced image.
The image enhancement and acquisition processes are parallelized, significantly reducing memory usage and processing latency, generating continuous, natural, and detailed images to meet real-time detection requirements and improve detection efficiency and accuracy.
Smart Images
Figure CN121998835A_ABST
Abstract
Description
Technical Field
[0001] This invention relates primarily to the field of semiconductor technology, and specifically to a real-time image enhancement method and system based on CD-SEM grating scanning block division. Background Technology
[0002] In semiconductor manufacturing processes, critical dimension scanning electron microscopy (CD-SEM) is a core instrument for chip linewidth measurement, defect detection, and morphology analysis. CD-SEM generates grayscale images by focusing an electron beam to scan the sample surface using a raster and collecting secondary or backscattered electron signals. However, due to factors such as sample material, charge effects, and electron beam noise, the acquired raw images often suffer from low overall contrast and blurred details, severely impacting the accuracy of subsequent automated measurement and defect identification.
[0003] Traditional image enhancement methods, such as Contrast-Limited Adaptive Histogram Equalization (CLAHE), divide the image into small blocks and perform histogram equalization on each block independently, while limiting excessive enhancement of local contrast. This aims to improve overall image contrast while highlighting local details. However, the traditional CLAHE algorithm is designed for fully acquired digital images. When faced with high-resolution images (e.g., 4096 pixels) generated by CD-SEM, its effectiveness diminishes. When applying the CLAHE algorithm directly to images (4096 pixels or higher), two main problems arise: 1. Huge memory overhead: The entire high-resolution image needs to be loaded into memory and processed in blocks, which places high demands on computer hardware.
[0004] 2. Processing delay: Processing can only begin after the entire image has been scanned and transmitted, which cannot meet the real-time requirements of online detection.
[0005] CD-SEM itself operates by performing raster scanning line by line and block by block. Existing technology treats image acquisition and image enhancement as two separate, sequential stages, which is inefficient and consumes a lot of memory. Summary of the Invention
[0006] To address the technical problems existing in the prior art, this invention provides a high-efficiency, low-memory-occupancy real-time image enhancement method and system based on CD-SEM raster scanning block segmentation.
[0007] To solve the above-mentioned technical problems, the technical solution proposed by this invention is as follows: A real-time image enhancement method based on CD-SEM raster scan segmentation includes the following steps: S1. Based on the preset image processing block size and overlapping area parameters, control the electron beam of CD-SEM to scan the area to be tested along the grating scanning path to obtain an image acquisition block; wherein the scanning path causes adjacent scanning physical blocks to have overlapping areas in the horizontal and / or vertical directions; after each image acquisition block is scanned, the data of the image acquisition block is output. S2. Obtain the image acquisition block output in step S1, and sequentially perform local histogram statistics, cropping and equalization processing on the current image acquisition block to generate the corresponding enhanced image processing block and output it. S3. Receive the enhanced image processing block output from step S2 in real time, and stitch the current image processing block with the stitched image; wherein, for pixels that do not overlap between the current image processing block and adjacent stitched image processing blocks, stitching is performed directly; for pixels in overlapping areas, a weighted average algorithm is used for grayscale fusion to generate the final overall enhanced image.
[0008] Preferably, in step S2, the specific process of cutting is as follows: Set a clipping limit (CL) for each local histogram; Iterate through each gray level of the local histogram. If the number of pixels at a gray level exceeds the cropping limit CL, then crop out the excess portion. The total number of cropped pixels is evenly redistributed across all gray levels to obtain the cropped and redistributed histogram.
[0009] Preferably, the formula for calculating the cutting limit CL is:
[0010] in This is an adjustable parameter.
[0011] Preferably, in step S2, the specific process of the equalization process is as follows: Calculate the cumulative distribution function (CDF) based on the histogram obtained after clipping. Based on the cumulative distribution function (CDF), a mapping function is constructed to obtain new grayscale values.
[0012] Preferably, the mapping function The formula is:
[0013] Let k be the mapping function, representing the new grayscale value after mapping the grayscale value k. The minimum value in CDF; N is the total number of pixels in the image block; L is the maximum number of gray levels; This represents the cumulative distribution function value for a gray value of k.
[0014] Preferably, in step S3, the weighted average algorithm is as follows: For any pixel A within the overlapping region, its grayscale value after equalization of the image patch on the left or top side is... The grayscale value of the image patch after equalization on the right or bottom side is The new grayscale value after pixel fusion Specifically:
[0015] in The distance of pixel A from the center of the image block on the left or top side. This represents the distance of pixel A from the center of the image block to the right or bottom.
[0016] Preferably, in step S1, the step size of the scanning path is aligned with or is an integer multiple of the size of a preset image processing block.
[0017] The present invention also discloses a computer program product, comprising a computer program that, when executed by a processor, performs the steps of the method described above.
[0018] The present invention further discloses a computer-readable storage medium having a computer program stored thereon, the computer program executing the steps of the method described above when run by a processor.
[0019] The present invention also discloses a real-time image enhancement system based on CD-SEM raster scan blocks, including a memory and a processor connected to each other. The memory stores a computer program, which executes the steps of the method described above when run by the processor.
[0020] Compared with the prior art, the advantages of the present invention are as follows: The core idea of the method of this invention is: "scanning is processing, acquisition is enhancement", which can parallelize the image enhancement process and the image acquisition process, significantly reduce memory usage and processing latency, and achieve real-time and efficient image enhancement.
[0021] This invention breaks down the massive image processing task into miniature tasks that are synchronized with scanning, enabling parallel pipeline operations for acquisition and processing. It eliminates the time spent waiting for the entire image to be scanned, greatly improves the image output speed, meets the requirements for real-time detection, and has the advantages of high efficiency and low latency.
[0022] The system of this invention does not require allocating a cache for the entire high-resolution image, but only needs to process the current and a few adjacent small blocks, which greatly reduces the demand for computer memory, making a standard industrial computer sufficient for high-resolution imaging scenarios.
[0023] The method of this invention makes full use of the working characteristics of CD-SEM hardware itself, without requiring any modification to the core electro-optical system, and can be easily implemented on existing CD-SEM equipment through software upgrades or the addition of dedicated processing boards.
[0024] This invention effectively eliminates the block effect common in traditional block processing through careful overlapping region design and fusion algorithm, resulting in a final image that is visually continuous, natural, and rich in detail. Attached Figure Description
[0025] Figure 1 This is a flowchart of an embodiment of the real-time image enhancement method of the present invention.
[0026] Figure 2 This is a diagram illustrating an embodiment of the real-time image enhancement system of the present invention in a specific application.
[0027] Figure 3 This is a schematic diagram of grating scanning in this invention.
[0028] Figure 4 This is a schematic diagram of the overlapping area between adjacent image blocks in this invention. Detailed Implementation
[0029] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.
[0030] like Figure 1 As shown, the real-time image enhancement method based on CD-SEM raster scan blocks provided in this embodiment of the invention includes the following steps: S1. Based on the preset image processing block size and overlapping area parameters, control the electron beam of CD-SEM to scan the area to be tested along the grating scanning path to obtain the image acquisition block; Specifically, this invention fully utilizes the characteristic of CD-SEM raster scanning (from left to right, from top to bottom) that naturally generates image block sequences. First, the block size (e.g., 64 pixels) is processed according to the algorithm. The scan step is set or fine-tuned using 64 pixels to align the physical scan block with the algorithm processing block in size or make them integer multiples of each other. To avoid boundary marks (block effect) caused by independent processing between blocks, a pixel overlap band is set between adjacent scan blocks when setting the scan path.
[0031] Specifically, after the operator sets the scanning parameters (such as total field of view and resolution) and CLAHE parameters (block size and number of overlapping pixels) in the software interface, the system logically divides the entire scanning area into a series of continuous image acquisition blocks. During scanning, the electron beam moves strictly along this block path, such as... Figure 3As shown. The key is that, to ensure subsequent stitching quality, the scan path is set to maintain a preset width of pixel overlap (e.g., 8 pixels) between adjacent scan blocks in both the horizontal and vertical directions, as shown. Figure 4 As shown. In this way, once the scanning of each image acquisition block is completed, the data of that block (i.e., the original grayscale image data) is immediately output and sent to the next step for processing, without waiting for the entire image to be scanned.
[0032] S2. Obtain the image acquisition block output in step S1, and sequentially perform local histogram statistics, cropping and equalization processing on the current image acquisition block to generate the corresponding enhanced image processing block and output it. Histogram statistics: For the current scanned image block, iterate through all its pixels and count the number of pixels appearing at each gray level (e.g., 0-255) to form the local histogram of the sub-block.
[0033] Cropping: To avoid over-amplification of small amounts of noise in uniform regions, local histograms need to be cropped. Specifically, a cropping limit CL is first set for each local histogram; then, the local histogram is iterated, and if the number of pixels at a certain gray level exceeds the cropping limit CL, the excess is "cropped". The total number of cropped pixels is then evenly redistributed across all gray levels, resulting in a cropped and redistributed histogram.
[0034] The cutting limit CL is usually calculated using the following formula:
[0035] in These are adjustable parameters, such as 2.0 or 3.0; Equalization: Based on the histogram after "cropping and redistribution", calculate its cumulative distribution function (CDF). The k-th value of the CDF is equal to the sum of the number of pixels at the first k gray levels in the histogram. The calculation formula is as follows:
[0036] k is the current gray level ( , (Maximum gray level); It is the number of pixels at gray level i after cropping and redistribution; Next, based on the cumulative distribution function (CDF), a mapping function is constructed to map and obtain new grayscale values. This mapping function "spreads out" the originally concentrated grayscale levels within an image patch, making dark areas darker and bright areas brighter, thereby greatly enhancing the contrast of that local area and making details clearer. The mapping function formula is as follows:
[0037] Let k be the mapping function, representing the new grayscale value after mapping the grayscale value k. The minimum value in CDF; N is the total number of pixels in the image block; L is the maximum number of gray levels; This represents the cumulative distribution function value for a gray value of k.
[0038] Finally, the above mapping function is applied to perform grayscale transformation on the current image patch, and the enhanced image patch is output. S3. Receive the enhanced image processing block output from step S2 in real time, and stitch the current image processing block with the stitched image to combine the enhanced image processing blocks into a complete image in an orderly manner.
[0039] To avoid block boundary effects caused by direct stitching of independently processed images, an overlapping region fusion strategy is adopted. When an enhanced image processing block is received, and it is fused with the already stitched portion of the image, only the pixels in the non-overlapping areas are directly stitched together, filling the corresponding positions in the final image. For overlapping areas, a weighted average fusion algorithm is used. Specifically, taking the horizontal overlapping area as an example, such as... Figure 4 As shown, for any pixel A within the overlapping area, let its distance from the center of the left image patch be p, and its distance from the center of the right image patch be q; the grayscale value of point A after equalization in the left image patch is... The grayscale value of the image patch on the right side of point A after equalization is The new grayscale value after fusion The formula is shown below:
[0040] Similarly, the merging of adjacent blocks vertically is done in a similar way. This merging achieves a smooth transition between blocks, effectively eliminating visual boundary marks and ensuring the global consistency and naturalness of the final image.
[0041] The real-time image enhancement method based on CD-SEM raster scanning block segmentation of the present invention has the following technical advantages: 1. Block-based scanning strategy: Instead of waiting for the entire image to be scanned, the feature of CD-SEM raster scanning is utilized to scan in preset "scan blocks". Once a block is scanned, processing is triggered immediately, thereby achieving "scanning, processing and output simultaneously". This fundamentally eliminates the waiting time for the entire image and meets the stringent requirements of high throughput in semiconductor testing.
[0042] 2. Overlapping Blocking and Region Fusion Mechanism: Overlapping sub-blocks are divided, and weighted fusion is performed in the overlapping areas. This avoids the gray-scale jumps (block artifacts) that occur at the block boundaries in traditional CLAHE, ensuring the overall smoothness and consistency of the final stitched image and improving the accuracy of subsequent measurements.
[0043] 3. Hardware-friendly streaming architecture: The entire method is designed as a continuous pipeline, with data flowing sequentially through the segmentation, processing, and fusion modules like water. This greatly reduces the reliance on large-capacity, high-bandwidth memory, making the algorithm very suitable for embedded deployment on FPGAs or dedicated DSPs, meeting the stringent requirements of CD-SEM devices for real-time performance, size, and power consumption.
[0044] To better understand the above technical solutions, the following will provide a detailed explanation of the technical solutions in conjunction with the accompanying drawings and specific implementation methods.
[0045] The overall flowchart of this invention is as follows: Figure 1 As shown, its core is a pipeline processing system synchronized with CD-SEM scanning, mainly including modules for "scanning segmentation," "histogram statistics, cropping, and equalization," and "image stitching and fusion." The module composition diagram is shown below. Figure 2 As shown.
[0046] This invention takes the detection of a polysilicon gate structure on a silicon wafer as an example, and specifically includes the following steps: 1. Initialization settings: 1.1) The operator places the sample into the CD-SEM sample chamber, evacuates the vacuum, and positions it in the area to be tested.
[0047] 1.2) Set the scanning parameters in the software interface: total field of view is 10μm. 10μm, image resolution 4096 4096 pixels.
[0048] 1.3) Set the CLAHE parameter: block size is 64 64 pixels, cropping limit of 2.0, overlap area of 8 pixels.
[0049] 1.4) The system logically divides the total field of view into 64 segments based on the CLAHE block size. 64 (i.e., 4096 / 64) image acquisition blocks. The scanning electron beam will scan these blocks sequentially along the raster path (from left to right, from top to bottom).
[0050] 2. Scanning and processing workflow: 2.1) The scanning control module drives the electron beam to move to the position of the first image acquisition block (located in the upper left corner) and begins raster scanning to divide the image into blocks. As the scanning progresses, the secondary electron signal is received by the detector and converted into a digital image signal, forming the first image acquisition block. The data of this block is immediately sent to the "histogram statistics, cropping, and equalization" module.
[0051] 2.2) The "Histogram Statistics, Cropping, and Equalization" module receives the first image acquisition block and processes it in real time. It constructs a grayscale histogram for the block, performs cropping and equalization calculations, applies grayscale transformations, and outputs the first enhanced image processing block.
[0052] 2.3) The “Image Stitching and Fusion” module receives the first image processing block. Since it is the starting block and there are no adjacent blocks to be merged, it is directly placed at the (0,0) coordinate position of the final image.
[0053] 2.4) Simultaneously, the electron beam of the CD-SEM has moved to the second image acquisition block (immediately to the right of the first block). Importantly, the scan range of this block includes an 8-pixel overlap with the first block in the horizontal X direction. After scanning is complete, the process in 2.2) is repeated.
[0054] 2.5) When placing the second processing block, the "Image Stitching and Fusion" module identified an overlap between its left side and the first block. The overlap and fusion module was activated, performing a linear weighted average fusion on the 8-pixel wide vertical strip.
[0055] 2.6) Repeat the above process until all blocks in the first row have been processed. Then the electron beam moves to the starting position of the second row, where the first block of the second row also overlaps with the corresponding block in the first row in the vertical Y direction, and performs a similar fusion process.
[0056] 2.7) This "scan-process-stitch" pipeline process continues. Before the scan ends, the display buffer of the image stitching module has already displayed most of the enhanced image, with only the bottom right corner possibly still being filled. The operator can observe almost in real time that a high-contrast, detailed image is being "drawn".
[0057] This embodiment demonstrates that the method of the present invention can effectively achieve real-time, high-quality enhancement of CD-SEM images, greatly improving the efficiency of semiconductor detection and user experience.
[0058] The core idea of the method of this invention is: "scanning is processing, acquisition is enhancement", which can parallelize the image enhancement process and the image acquisition process, significantly reduce memory usage and processing latency, and achieve real-time and efficient image enhancement.
[0059] This invention breaks down the massive image processing task into miniature tasks that are synchronized with scanning, enabling parallel pipeline operations for acquisition and processing. It eliminates the time spent waiting for the entire image to be scanned, greatly improves the image output speed, meets the requirements for real-time detection, and has the advantages of high efficiency and low latency.
[0060] The system of this invention does not require allocating a cache for the entire high-resolution image, but only needs to process the current and a few adjacent small blocks, which greatly reduces the demand for computer memory, making a standard industrial computer sufficient for high-resolution imaging scenarios.
[0061] The method of this invention makes full use of the working characteristics of CD-SEM hardware itself, without requiring any modification to the core electro-optical system, and can be easily implemented on existing CD-SEM equipment through software upgrades or the addition of dedicated processing boards.
[0062] This invention effectively eliminates the block effect common in traditional block processing through careful overlapping region design and fusion algorithm, resulting in a final image that is visually continuous, natural, and rich in detail.
[0063] The present invention also discloses a computer program product, comprising a computer program that, when executed by a processor, performs the steps of the method described above.
[0064] The present invention further discloses a computer-readable storage medium having a computer program stored thereon, the computer program executing the steps of the method described above when run by a processor.
[0065] The present invention also discloses a real-time image enhancement system based on CD-SEM raster scan blocks, including a memory and a processor connected to each other. The memory stores a computer program, which executes the steps of the method described above when run by the processor.
[0066] The products, media, and systems of the present invention, corresponding to the methods described above, also possess the advantages described above.
[0067] The present invention can implement all or part of the processes in the methods of the above embodiments, or it can be implemented by hardware related to computer program instructions. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by a processor, it can implement the steps of the above method embodiments. The computer program includes computer program code, which can be in the form of source code, object code, executable file, or some intermediate form. The computer-readable storage medium includes: any entity or device capable of carrying computer program code, recording media, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. The memory is used to store computer programs and / or modules. The processor implements various functions by running or executing the computer programs and / or modules stored in the memory, and by calling data stored in the memory. The memory may include high-speed random access memory, as well as non-volatile memory, such as hard disks, RAM, plug-in hard disks, smart media cards (SMC), secure digital (SD) cards, flash cards, at least one disk storage device, flash memory device, or other volatile solid-state storage devices.
[0068] Explanation of related terms: CD-SEM: Scanning electron microscope for critical dimensions; CLAHE: A contrast-limited adaptive histogram equalization algorithm; FPGA: Field Programmable Gate Array; DSP: Digital Signal Processor.
[0069] The above are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should be considered within the scope of protection of the present invention.
Claims
1. A real-time image enhancement method based on CD-SEM raster scanning block segmentation, characterized in that, Including the following steps: S1. Based on the preset image processing block size and overlapping area parameters, control the electron beam of CD-SEM to scan the area to be tested along the grating scanning path to obtain an image acquisition block; wherein the scanning path causes adjacent scanning physical blocks to have overlapping areas in the horizontal and / or vertical directions; after each image acquisition block is scanned, the data of the image acquisition block is output. S2. Obtain the image acquisition block output in step S1, and sequentially perform local histogram statistics, cropping and equalization processing on the current image acquisition block to generate the corresponding enhanced image processing block and output it. S3. Receive the enhanced image processing block output from step S2 in real time, and stitch the current image processing block with the stitched image; wherein, for pixels that do not overlap between the current image processing block and adjacent stitched image processing blocks, stitching is performed directly; for pixels in overlapping areas, a weighted average algorithm is used for grayscale fusion to generate the final overall enhanced image.
2. The real-time image enhancement method based on CD-SEM raster scanning block segmentation according to claim 1, characterized in that, In step S2, the specific process of cutting is as follows: Set a clipping limit (CL) for each local histogram; Iterate through each gray level of the local histogram. If the number of pixels at a gray level exceeds the cropping limit CL, then crop out the excess portion. The total number of cropped pixels is evenly redistributed across all gray levels to obtain the cropped and redistributed histogram.
3. The real-time image enhancement method based on CD-SEM raster scanning block segmentation according to claim 2, characterized in that, The formula for calculating the cropping limit CL is as follows: in This is an adjustable parameter.
4. The real-time image enhancement method based on CD-SEM raster scanning block segmentation according to claim 1, 2, or 3, characterized in that, In step S2, the specific process of the equalization process is as follows: Calculate the cumulative distribution function (CDF) based on the histogram obtained after clipping. Based on the cumulative distribution function (CDF), a mapping function is constructed to obtain new grayscale values.
5. The real-time image enhancement method based on CD-SEM raster scanning block segmentation according to claim 4, characterized in that, Mapping function The formula is: Let k be the mapping function, representing the new grayscale value after mapping the grayscale value k. The minimum value in CDF; N is the total number of pixels in the image block; L is the maximum number of gray levels; This represents the cumulative distribution function value for a gray value of k.
6. The real-time image enhancement method based on CD-SEM raster scanning block segmentation according to claim 1, 2, or 3, characterized in that, In step S3, the weighted average algorithm is as follows: For any pixel A within the overlapping region, its grayscale value after equalization of the image patch on the left or top side is... The grayscale value of the image patch after equalization on the right or bottom side is The new grayscale value after pixel fusion Specifically: in The distance of pixel A from the center of the image block on the left or top side. This represents the distance of pixel A from the center of the image block to the right or bottom.
7. The real-time image enhancement method based on CD-SEM raster scanning block segmentation according to claim 1, 2, or 3, characterized in that, In step S1, the step size of the scanning path is aligned with or is an integer multiple of the size of the preset image processing block.
8. A computer program product, comprising a computer program, characterized in that, The computer program is executed by the processor to perform the steps of the method as described in any one of claims 1-7.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that, The computer program, when run by a processor, performs the steps of the method as described in any one of claims 1-7.
10. A real-time image enhancement system based on CD-SEM raster scan segmentation, comprising an interconnected memory and a processor, wherein the memory stores a computer program, characterized in that, The computer program, when run by a processor, performs the steps of the method as described in any one of claims 1-7.