Dic system-based roi region identification method and system

Abstract

The application discloses a ROI region identification method and system based on a DIC system, and belongs to the technical field of optical measurement and digital image analysis. The application aims to solve the problems of large subjective error, low efficiency and hindering process automation caused by manual selection of a region of interest (ROI) in the existing DIC technology. The application uses the high-frequency texture characteristics of a speckle image, constructs a second-order derivative feature surface by calculating the second-order derivative of the image gray scale, carries out binaryzation processing on the feature surface to obtain an initial binary mask, carries out morphological correction, connected region analysis and area filtering on the initial mask, screens out an effective connected region, and extracts the boundary of the effective connected region and converts it into a polygon representation. The application realizes the objectivity and high repeatability of ROI identification, greatly improves the processing efficiency, and enables the DIC technology to realize the full-process automation from image input to result output.

Description

Technical Field

[0001] This invention belongs to the field of optical measurement and digital image correlation (DIC) technology, specifically relating to a method and system for ROI region identification based on a DIC system, so as to achieve automatic, accurate and repeatable identification of ROI regions in DIC analysis. Background Technology

[0002] Digital image correlation (DIC) is a non-contact, high-precision optical measurement method widely used in materials mechanics, structural engineering, biomedicine, and other fields for full-field deformation and strain measurement. Its basic process involves comparing speckle images of an object's surface before and after deformation to calculate the displacement and strain fields. In DIC analysis, the first and crucial step is defining the region of interest (ROI), which is the effective area in the image for which correlation calculations are performed. Precise ROI definition greatly improves computational efficiency and accuracy, eliminating interference from irrelevant backgrounds.

[0003] However, in existing technologies, the definition of ROI heavily relies on manual selection, which has the following significant drawbacks: Large human error and poor repeatability: Different operators, and even the same operator at different times, may select ROIs with subjective differences in position, size and boundary, resulting in poor data comparability.

[0004] Extremely inefficient and unsuitable for batch analysis: Manually adjusting the ROI is extremely time-consuming in experimental scenarios with multiple samples and multiple operating conditions.

[0005] Obstacles to full-process automation: Manual ROI definition is a "breakpoint" in the DIC analysis process that requires manual intervention, making it impossible to form a closed-loop automated process of "image acquisition → DIC calculation → post-processing".

[0006] Furthermore, although there are various general image segmentation algorithms, they are not directly applicable to the task of automatic ROI identification in DIC speckle images: classic edge detection is sensitive to individual edges, but DIC dense speckle will produce a large number of fragmented edge responses; global thresholding segmentation relies on the bimodal distribution of the foreground and background gray-level histograms, while the DIC speckle region contains bright and dark spots, and its overall gray-level mean may be similar to the background; region growing algorithms are extremely sensitive to the initial seed point; although deep learning segmentation models have high accuracy, they require a large amount of accurately labeled data for training.

[0007] Therefore, developing a technology that can automatically, accurately, and consistently identify ROI regions in DIC analysis, and that is theoretically compatible with the characteristics of speckle images and computationally efficient, has become an urgent need to improve the reliability, automation level, and engineering applicability of DIC methods. Summary of the Invention

[0008] To address the aforementioned technical problems, this invention provides a method and system for ROI region identification based on a DIC system, thereby overcoming the technical defects of existing technologies, such as strong reliance on manual intervention, poor repeatability, and low efficiency.

[0009] This invention constructs an image analysis process centered on "second derivative feature surfaces" and utilizes the unique high-frequency texture features of speckle images to achieve automatic identification and boundary extraction of ROI regions.

[0010] Specifically, this invention utilizes the characteristic that the effective areas of DIC analysis are typically characterized by high-contrast speckle patterns formed by spraying or natural processes. These speckles manifest as frequent, localized abrupt changes in grayscale values ​​on the image. The second derivative (Laplacian operator) of an image is extremely sensitive to the rate of grayscale change, responding significantly to edges, corners, and textured areas. Therefore, the absolute value or energy of the Laplacian in speckle regions is significantly higher than that of a uniform background. By calculating and analyzing the "second derivative feature surface" of the entire image, the boundary between the ROI and the background can be objectively and quantitatively defined.

[0011] All image processing steps are completed in this analysis layer, which then drives subsequent DIC calculations. By inputting a speckle image and converting it into a grayscale matrix, the Laplacian matrix is ​​calculated and a second-order derivative feature surface is constructed. Binarization is performed to obtain an initial mask, and then effective connected regions are selected through morphological correction and connected region analysis. Finally, the boundaries are extracted and converted into polygon representations.

[0012] Thus, this invention achieves complete automation from image input to ROI boundary output, significantly improving the automation level, repeatability, and processing efficiency of DIC analysis.

[0013] According to one aspect of the present invention, a method for ROI region identification based on a DIC system is provided, comprising the following steps: Input the speckle image to be analyzed, convert the speckle image into a grayscale image, and obtain the grayscale matrix corresponding to the grayscale image.

[0014] The second derivative is calculated by applying the Laplacian operator to the grayscale matrix to obtain the second derivative matrix, and the second derivative feature surface is constructed based on the second derivative matrix.

[0015] The second derivative feature surface is binarized to obtain an initial binary mask.

[0016] The initial binary mask is morphologically corrected, and the morphologically corrected mask is subjected to connected component analysis. Valid connected components are then selected from the analysis results.

[0017] Extract the boundary of each of the effective connected regions and transform the boundary into a polygon representation of the ROI region.

[0018] Another aspect of the present invention provides a Region of Interest (ROI) identification system based on a DIC system, comprising: The image input module is used to input the speckle image to be analyzed, convert the speckle image into a grayscale image, and obtain the grayscale matrix corresponding to the grayscale image.

[0019] The feature calculation module is used to apply the Laplacian operator to the grayscale matrix to calculate the second derivative, obtain the second derivative matrix, and construct the second derivative feature surface based on the second derivative matrix.

[0020] The ROI segmentation and optimization module is used to binarize the second derivative feature surface to obtain an initial binary mask, perform morphological correction on the initial binary mask, and perform connected component analysis on the morphologically corrected mask to select effective connected components from the analysis results.

[0021] The contour generation module is used to extract the boundary of each of the effective connected regions and convert the boundary into a polygon representation of the ROI region.

[0022] The process control module is used to coordinate the sequential execution of the image input module, the feature calculation module, the ROI segmentation and optimization module, and the contour generation module, and output the polygon representation.

[0023] Compared with the prior art, the present invention has the following beneficial effects: Completely eliminates human error and achieves excellent repeatability: ROI recognition is based entirely on the objective mathematical features of the image (second derivative). The same image, processed by any system at any time, will yield completely consistent ROI boundaries as long as the parameters are consistent.

[0024] Achieving high-speed, fully automated batch processing: This method requires absolutely no manual intervention and can process single images instantly. When embedded with automated scripts, it can perform frame-by-frame ROI identification and adaptive updates for massive image sequences.

[0025] Achieving end-to-end automation: As a reliable automation front-end, this invention enables true end-to-end automation of the entire DIC analysis process (acquisition-identification-calculation-output).

[0026] High flexibility and robustness: By adjusting the threshold, morphological operation parameters and indentation, this method can adapt to various experimental images with different contrasts, lighting conditions and speckle densities. Attached Figure Description

[0027] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0028] Figure 1 This is a schematic diagram of the ROI region identification system based on the DIC system provided in an embodiment of the present invention; Figure 2 This is a flowchart of the ROI region identification method based on the DIC system provided in the embodiments of the present invention; Figure 3a Original speckle image; Figure 3b A visualization of the surface containing the calculated absolute value of the second derivative; Figure 3c This is the binary mask image of the ROI obtained after thresholding and morphological processing. Figure 3d This is a rendering of the smoothed ROI boundary contours automatically extracted and overlaid on the original image. Figure 3e The final map showing the effective ROI regions determined for DIC analysis; Figure 4a , Figure 4b This is a comparative diagram showing the automatic identification method of the present invention and the traditional manual selection method for determining the ROI region. Figure 5 This is a schematic diagram illustrating the ROI identification of samples with different speckles and shapes according to the present invention.

[0029] Terminology Explanation To facilitate understanding of the technical solution of this invention, some technical terms appearing in the specification are explained as follows: Digital Image Correlation (DIC): A non-contact optical measurement technique that calculates displacement and strain fields by comparing speckle images of an object's surface before and after deformation.

[0030] Region of Interest (ROI): The effective image region that needs to be calculated in DIC analysis.

[0031] The Laplacian Operator is a second-order differential operator used to detect regions of rapidly changing grayscale values ​​in an image. It is sensitive to edge and texture features.

[0032] Second derivative matrix: The matrix obtained after applying the Laplacian operator to a grayscale image, where the value of each pixel represents the second derivative response of the grayscale at that point.

[0033] Second derivative feature surface: A feature map constructed based on the second derivative matrix to characterize the degree of drastic changes in local texture of an image, and is the basis for subsequent thresholding segmentation.

[0034] Opening operation: A morphological image processing operation that involves erosion followed by dilation, used to eliminate noise points and burrs in an image.

[0035] Closing operation: A morphological image processing operation that involves dilation followed by erosion, used to fill small holes and gaps in an image.

[0036] Connected component analysis: The process of identifying and labeling interconnected pixel regions in a binary image.

[0037] Polygon approximation: Simplifying the sequence of outline pixels into a polygon representation with fewer vertices. Commonly used algorithms include the Douglas-Peucker algorithm. Detailed Implementation

[0038] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0039] The embodiments of this invention are written in a progressive manner. Example 1: System Architecture

[0040] like Figure 1 As shown, this invention provides a ROI region recognition system based on a DIC system. Its core lies in constructing an image analysis process with "second derivative feature surface" as the core, and utilizing the high-frequency texture features unique to speckle images to achieve automatic recognition and boundary extraction of ROI regions.

[0041] The system consists of five core modules that work together to achieve a complete closed loop from image input to ROI boundary output: The image input module is used to input the speckle image to be analyzed, convert the speckle image into a grayscale image, and obtain the grayscale matrix corresponding to the grayscale image. The input of this module includes the original speckle image file (such as BMP, JPEG, PNG, etc.).

[0042] The feature calculation module is used to apply the Laplacian operator to the grayscale matrix to calculate the second derivative, obtaining the second derivative matrix, and constructing a second derivative feature surface based on the second derivative matrix. The core of this module is: applying a discrete Laplacian convolution kernel to the grayscale matrix to obtain the second derivative matrix, taking the absolute value to obtain the second derivative intensity surface, and then calculating the standard deviation within the local neighborhood window of each pixel to generate the final second derivative feature surface.

[0043] It should be noted that the second-derivative feature surface is not a simple transformation of the original grayscale image, but rather, through standard deviation calculation within a local window, it transforms the second-derivative response of each pixel into a statistical measure of the degree of texture fluctuation in its surrounding neighborhood. Due to frequent abrupt changes in grayscale values ​​in speckle-dense regions, the standard deviation within the local window is significantly higher than that in uniform background regions. Therefore, the second-derivative feature surface can effectively amplify the difference between speckle regions and background regions, providing a good numerical basis for subsequent thresholding.

[0044] The ROI segmentation and optimization module is used to binarize the second-derivative feature surface to obtain an initial binary mask, perform morphological correction on the initial binary mask, and conduct connected component analysis on the morphologically corrected mask to filter out effective connected regions from the analysis results. This module generates the initial mask by comparing the feature surface with a global threshold, optimizes the mask shape through opening and closing operations, and removes noisy regions through area filtering.

[0045] The contour generation module is used to extract the boundary of each effective connected region and transform the boundary into a polygonal representation of the ROI region. This module extracts the contour pixel set of each effective connected region and simplifies the dense sequence of contour points into a polygonal representation with fewer vertices using a polygon approximation algorithm (such as the Douglas-Peucker algorithm).

[0046] The flow control module coordinates the sequential execution of the image input module, the feature calculation module, the ROI segmentation and optimization module, and the contour generation module, and outputs the polygon representation. This module provides a unified API interface and can be embedded at the forefront of the DIC calculation software flow.

[0047] The five functional modules mentioned above work together to form a complete closed-loop system. After the user inputs a speckle image, the image input module completes grayscale conversion; the feature calculation module generates a second-derivative feature surface; the ROI segmentation and optimization module completes threshold segmentation and morphological correction to select effective connected regions; the contour generation module extracts the boundaries and converts them into polygon representations; and the flow control module coordinates the sequential execution of each module and outputs the results.

[0048] All modules communicate with each other through well-defined APIs, ensuring the system's scalability and stability. Example 2: General Process

[0049] like Figure 2 As shown, the present invention provides a ROI region identification method based on a DIC system, comprising the following steps S1 to S6: S1: Image Input and Preprocessing Input the speckle image to be analyzed, convert the speckle image into a grayscale image, and obtain the grayscale matrix corresponding to the grayscale image. If the original image is a color image, convert it into a grayscale image using a standard grayscale conversion formula (e.g., Y = 0.299R + 0.587G + 0.114B).

[0050] S2: Calculate the characteristic surface of the second derivative. The second derivative is calculated by applying the Laplacian operator to the grayscale matrix to obtain the second derivative matrix, and the second derivative feature surface is constructed based on the second derivative matrix.

[0051] Specifically, the process includes the following sub-steps: applying a discrete Laplacian convolution kernel to the grayscale matrix to obtain a second derivative matrix; taking the absolute value of the second derivative matrix to obtain a second derivative intensity surface; calculating the statistical characteristics (standard deviation) of the second derivative intensity surface within the local neighborhood window of each pixel; and generating a second derivative feature surface based on the statistical characteristics. This surface characterizes the degree of drastic change in speckle texture in the image, with the values ​​in the speckle region being significantly higher than those in the uniform background region.

[0052] The general engineering window size range is [3×3 pixels, 25×25 pixels], preferably 11×11 pixels.

[0053] S3: Automatic threshold segmentation of ROI regions The second-order derivative feature surface is binarized to obtain an initial binary mask. Specifically, the binarization process involves classifying pixel values ​​in the second-order derivative feature surface using a global threshold, and identifying pixels with feature values ​​greater than the global threshold as potential Regions of Interest (ROIs).

[0054] The global threshold is adaptively determined based on the global numerical distribution characteristics of the second derivative feature surface. Specifically, the global mean and global maximum value of the second derivative feature surface are calculated, and the global threshold is calculated based on a preset percentage of the global mean and the global maximum value.

[0055] As a preferred implementation method, a percentage coefficient is set. The general engineering range is [0.25, 0.55], the preferred value is 0.4, and the threshold T = global mean. + *(Global maximum value) -Global Mean ).

[0056] S4: Morphological Correction The initial binary mask is morphologically modified to obtain the morphologically modified mask.

[0057] Specifically, the process includes the following sub-steps: performing an opening operation (erosion followed by dilation) on the initial binary mask to obtain an opening operation mask to eliminate noise points and burrs; performing a closing operation (dilation followed by erosion) on the opening operation mask to obtain a closing operation mask to fill small holes and gaps; and using the closing operation mask as the morphologically corrected mask.

[0058] As a preferred embodiment, a boundary indentation operation may also be included: performing a boundary indentation operation on the closing operation mask, i.e., performing morphological erosion on the closing operation mask at a specific pixel distance, to obtain an eroded mask, which is then used as the morphologically corrected mask. This aims to shrink the ROI boundary inward from the potentially unstable physical sample edge, ensuring that the defined ROI is entirely within a stable and clear speckle field.

[0059] The indentation pixel count, dpixel, can be determined based on the spatial resolution, PPI (pixels per mm), and the physical indentation distance, Dphy. ; The physical indentation distance Dphy ranges from [sample size X*1%, sample size X*10%], with a preferred value of sample size X*5%. This indentation operation aims to shrink the ROI boundary inward from the edge of the potentially unstable physical sample, ensuring that the defined ROI is completely located within a stable and clear speckle field.

[0060] S5: Connectivity Analysis and Effective Region Selection Connectivity analysis is performed on the morphologically corrected mask, and valid connected regions are selected from the analysis results.

[0061] Specifically, the process includes the following sub-steps: performing connected component marking on the morphologically corrected mask to obtain one or more connected components; calculating the area of ​​each connected component; comparing the area of ​​each connected component with a preset minimum area threshold; removing connected components with an area smaller than the minimum area threshold (usually noise), and retaining the remaining connected components as the valid connected components.

[0062] The minimum area threshold ranges from [10 pixels to 240 pixels], with 120 pixels being the preferred value.

[0063] S6: Contour Extraction and Output Extract the boundary of each of the effective connected regions and transform the boundary into a polygon representation of the ROI region.

[0064] Specifically, the process includes the following sub-steps: for each effective connected region, extract its contour pixel set; use the contour pixel set as the boundary; perform polygon approximation on the contour pixel set (e.g., using the Douglas-Peucker algorithm) to obtain a polygon vertex sequence; use the polygon vertex sequence as the polygon representation. This polygon representation can be directly used to define the template region and search region for DIC calculation.

[0065] Example 3: Specific Application Case The method of the present invention will be described in detail using the speckle image of a material's mechanical tensile test as an example.

[0066] Image Input: Import the raw speckle image to be analyzed (e.g., Figure 3a As shown, the image format is 8-bit grayscale BMP, and the resolution is 1280×1024 pixels. Convert the image to grayscale and obtain the grayscale matrix.

[0067] Feature surface calculation: A 3×3 Laplacian convolution kernel is applied to the grayscale matrix to obtain the second derivative matrix. After taking the absolute value, the standard deviation of each pixel is calculated using a 21×21 pixel local window to generate the second derivative feature surface (e.g., ...). Figure 3b (As shown). In the characteristic surface, the values ​​of the speckle-dense regions are significantly higher than those of the uniform background regions.

[0068] Threshold segmentation: Calculate the global mean and global maximum of the feature surface. The global threshold is adaptively determined based on the global statistical properties of the second-derivative feature surface. As a specific implementation, the global mean of the second-derivative feature surface is calculated. And the global maximum value Max, by a preset percentage Calculate the global threshold In this embodiment Set the value to 0.3. Mark pixels with feature values ​​greater than T as ROIs and generate an initial binary mask (e.g., ...). Figure 3c (As shown).

[0069] Morphological correction: An opening operation (denoising) is performed on the initial mask using a 3×3 structuring element, followed by a closing operation (hole filling) to obtain an optimized mask. Optionally, a 12-pixel indentation is performed at the boundary to eliminate potential interference from the sample edges (e.g., Figure 3d (As shown).

[0070] Connected region filtering: Connected region labeling is performed on the optimized mask, and the area of ​​each connected region is calculated. A minimum area threshold of 200 pixels is set; connected regions with areas smaller than this threshold are removed, and the remaining connected regions are retained as valid connected regions.

[0071] Contour Output: Extract the contour pixel set for each valid connected region, approximate the polygon using the Douglas-Peucker algorithm, and set the approximation precision to epsilon = 0.005 × contour perimeter to obtain the polygon vertex sequence. Finally, output this polygon representation as the ROI boundary (e.g., ...). Figure 3e (As shown).

[0072] Figure 4a , Figure 4b The invention demonstrates a comparison between the automatic identification method of the present invention and the traditional manual selection method, showing that the method of the present invention has significant advantages in terms of efficiency and repeatability. Figure 5 The results of the ROI identification by the method of the present invention for samples with different speckle densities and different shapes are shown, and the effective ROI areas can be accurately identified. Example 4: Simulation Verification

[0073] To verify the quantitative accuracy of the method of the present invention, simulation tests were conducted using computer-generated Gaussian speckle images.

[0074] Generate a reference image with a size of 1280×1024 pixels, containing 15,000 speckle patterns with randomly distributed speckle radii of 2–4 pixels. The average gray level of the speckle region is 150–200, while the background region has a gray level of 100. An irregularly shaped Region of Interest (ROI) is pre-defined in the reference image to simulate the effective area of ​​the sample in actual DIC analysis; its pre-defined area is [missing information]. 489632 pixels.

[0075] The above-mentioned simulated image is processed according to the method of the present invention: S1: Image Input and Preprocessing: Input a simulated speckle image, convert it to a grayscale image, and obtain the grayscale matrix.

[0076] S2: Calculate the second derivative feature surface: Apply a 3×3 Laplacian convolution kernel to the grayscale matrix to calculate the second derivative, take the absolute value, and then use a 21×21 pixel local window to calculate the standard deviation to generate the second derivative feature surface.

[0077] S3: Automatic thresholding of ROI regions: Calculates the global mean μ and global maximum Max of the feature surface, and applies a preset percentage. =0.3 Calculate the global threshold Pixels with feature values ​​greater than T are identified as potential ROI regions, and an initial binary mask is generated.

[0078] S4: Morphological Correction and Connectivity Component Selection: The initial mask is subjected to opening and closing operations using a 3×3 structuring element, followed by a boundary indentation erosion operation with a distance of 12 pixels. Connectivity component marking is then performed on the corrected mask, eliminating connected components with an area less than 200 pixels, and retaining the remaining connected components as valid connected components.

[0079] S5: Contour Extraction and Output: Extract the outer contour of each effective connected region using the Douglas-Peucker algorithm ( The polygon approximation is performed using a formula equal to 0.005 × perimeter of the outline. The output is a sequence of polygon vertices representing the ROI boundary, and the measured area of ​​the identified ROI region is calculated. .

[0080] Verification results: After the above processing, the ROI area measurement value Compared with preset value The relative error is less than 1.8%, and the intersection-over-union (IoU) ratio is greater than 0.96. When the same simulated image is measured 10 times repeatedly, the standard deviation of the ROI area measurement is less than 0.2 pixels. 2 The standard deviation of the ROI boundary location is less than 0.4 pixels, and the coefficient of variation is less than 5×10. -7 .

[0081] The simulation results above demonstrate that the method of the present invention can achieve high-precision quantitative identification of ROI regions, with good measurement accuracy and repeatability, completely eliminating human error in traditional manual selection methods, and providing accurate and consistent ROI definitions for subsequent DIC calculations.

[0082] The scope of protection of this invention is not limited to the specific embodiments described above. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this invention, as well as the method for automatically defining the ROI region of DIC analysis by calculating the second derivative feature surface of speckle images, should be included within the scope of protection of this invention.

[0083] The foregoing has provided a detailed description of a ROI region identification method and system based on a DIC system provided in this application. The above description of the disclosed embodiments enables those skilled in the art to implement or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for ROI region identification based on a DIC system, characterized in that, Includes the following steps: Input the speckle image to be analyzed, convert the speckle image into a grayscale image, and obtain the grayscale matrix corresponding to the grayscale image; The second derivative is calculated by applying the Laplacian operator to the gray-level matrix to obtain the second derivative matrix, and a second derivative feature surface is constructed based on the second derivative matrix. The second derivative feature surface is binarized to obtain an initial binary mask; The initial binary mask is morphologically corrected, and the morphologically corrected mask is subjected to connected component analysis. Valid connected components are then selected from the analysis results. Extract the boundary of each of the effective connected regions and transform the boundary into a polygon representation of the ROI region.

2. The ROI region identification method based on the DIC system according to claim 1, characterized in that, The construction of the second-order derivative characteristic surface based on the second-order derivative matrix includes the following steps: Taking the absolute value of the second derivative matrix yields the second derivative intensity surface; Calculate the statistical characteristics of the second derivative intensity surface within the local neighborhood window of each pixel; The second derivative feature surface is generated based on the statistical characteristics.

3. The ROI region identification method based on the DIC system according to claim 2, characterized in that, The statistical feature is the standard deviation of the pixel values ​​within the local neighborhood window.

4. The ROI region identification method based on the DIC system according to claim 1, characterized in that, The binarization process specifically involves classifying the pixel values ​​in the second-order derivative feature surface using a global threshold.

5. The ROI region identification method based on the DIC system according to claim 4, characterized in that, The global threshold is adaptively determined based on the global numerical distribution characteristics of the second derivative feature surface.

6. The ROI region identification method based on the DIC system according to claim 5, characterized in that, The adaptive determination of the global threshold based on the global numerical distribution characteristics of the second derivative feature surface includes the following steps: Calculate the global mean and global maximum of the second derivative characteristic surface; The global threshold is calculated as a preset percentage of the global mean and the global maximum value.

7. The ROI region identification method based on the DIC system according to claim 1, characterized in that, The morphological correction of the initial binary mask includes the following steps: Perform an opening operation on the initial binary mask to obtain an opening operation mask; Perform a closing operation on the opening mask to obtain a closing mask; The closing operation mask is used as the morphologically modified mask.

8. The ROI region identification method based on the DIC system according to claim 7, characterized in that, The morphological correction of the initial binary mask further includes the following steps: Perform a boundary indentation operation on the closing operation mask to obtain an erosion mask; The erosion mask is used as the morphologically corrected mask.

9. The ROI region identification method based on the DIC system according to claim 1, characterized in that, The process of performing connected component analysis on the morphologically corrected mask and filtering out valid connected components from the analysis results includes the following steps: Perform connected component marking on the morphologically corrected mask to obtain one or more connected components; Calculate the area of ​​each of the connected regions; The area of ​​each connected region is compared with a preset minimum area threshold; Remove connected regions whose area is smaller than the minimum area threshold, and retain the remaining connected regions as the valid connected regions.

10. A Region of Interest (ROI) identification system based on a DIC system, characterized in that, include: The image input module is used to input the speckle image to be analyzed, convert the speckle image into a grayscale image, and obtain the grayscale matrix corresponding to the grayscale image; The feature calculation module is used to apply the Laplacian operator to the gray-level matrix to calculate the second derivative, obtain the second derivative matrix, and construct the second derivative feature surface based on the second derivative matrix; The ROI segmentation and optimization module is used to binarize the second derivative feature surface to obtain an initial binary mask, perform morphological correction on the initial binary mask, and perform connected component analysis on the morphologically corrected mask to filter out effective connected components from the analysis results. The contour generation module is used to extract the boundary of each effective connected region and convert the boundary into a polygon representation of the ROI region. The process control module is used to coordinate the sequential execution of the image input module, the feature calculation module, the ROI segmentation and optimization module, and the contour generation module, and output the polygon representation.

Images