Pressure analysis method of image data and related system thereof
By converting image data into grayscale images and using the K-means algorithm to cluster them, the problem of uncertain compressed locations in image data is solved, and accurate quantification of compressed areas in image data and improved computational efficiency are achieved.
Patent Information
- Application Number
- CN202410294835.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-03-14
- Publication Date
- 2025-09-26
AI Technical Summary
In existing computer-aided engineering (CAD) technology, the compressed locations of image data cannot be determined in advance, resulting in increased operational complexity and excessive consumption of server computing resources. Furthermore, existing software struggles to accurately output the compressed areas corresponding to each pressure range.
By converting image data into grayscale images and using the K-means algorithm to group each pixel, the compression state of the image is determined, and the compression analysis is performed in combination with the machine learning model.
It achieves accurate quantification of compressed areas in image data, reduces human estimation errors, and improves computational efficiency and data accuracy.
Smart Images
Figure CN120707900A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to a method for analyzing image data under pressure and a related system thereof, and more particularly to a method for analyzing image data under pressure and a related system thereof that can classify image pixels using a machine learning model. Background Art
[0002] Existing techniques for analyzing Computer Aided Engineering (CAE) data require determining the size of the compressed area within the image data. However, since the compressed locations within the image data cannot be determined in advance, the compressed locations must be determined before analyzing the compressed area and outputting the compressed data. This method increases operational complexity, and the repetitive calculation process can also overconsume server computing resources.
[0003] In addition, existing computer-aided engineering mechanical analysis software, such as Abaqus and Ansys, can perform mechanical analysis on nonlinear finite elements. However, Abaqus only provides the option to output the total compressed area and cannot output the content corresponding to each pressure range. It can only be estimated manually, and it is difficult to ensure the accuracy of the data.
[0004] Therefore, the existing technology needs to be improved. Summary of the Invention
[0005] Therefore, the present application provides a compression analysis method for image data and a related system thereof to quantify the compression analysis of image data.
[0006] The present invention discloses a compression analysis method for image data, comprising inputting an image; converting the image into a grayscale image; grouping each pixel of the grayscale image to obtain a plurality of image data groups; and determining a compression state of the image based on the plurality of image data groups.
[0007] An embodiment of the present application further discloses a compression analysis system for image data, comprising a processing device; and a memory device coupled to the processing device for storing a program code to instruct the processing device to execute a compression analysis process for an image, wherein the compression analysis process includes: inputting an image; converting the image into a grayscale image; grouping each pixel of the grayscale image to obtain a plurality of image data groups; and determining a compression state of the image based on the plurality of image data groups. BRIEF DESCRIPTION OF THE DRAWINGS
[0008] Figure 1 Shown is a schematic diagram of a pressure analysis system for image data according to an embodiment of the present application.
[0009] Figure 2Shown is a schematic diagram of a pressure analysis process according to an embodiment of the present application.
[0010] Figure 3 A schematic diagram showing the quantification of stress area of a microphone rubber according to an embodiment of the present application is shown.
[0011] Figure 4 Shown is a schematic diagram of grouping image data groups according to an embodiment of the present application.
[0012] Figure 5 Shown is a statistical diagram of the distribution of grayscale RGB values of an image before clustering using a K-means algorithm according to an embodiment of the present application.
[0013] Figure 6 Shown is a statistical diagram of the distribution of grayscale RGB values of an image after clustering using the K-means algorithm according to an embodiment of the present application.
[0014] Component number description
[0015] 10Compression Analysis System for Image Data
[0016] 102 processing device
[0017] 104 memory devices
[0018] 20 Image Data Compression Analysis Process
[0019] Steps 201-210 DETAILED DESCRIPTION
[0020] Please refer to Figure 1 , Figure 1 FIG1 is a schematic diagram of an image data pressure analysis system 10 according to an embodiment of the present application. The pressure analysis system 10 includes a processing device 102 and a memory device 104. The memory device 104 is coupled to the processing device 102 and stores a program code for instructing the processing device 102 to execute a pressure analysis process 20 for image data. The pressure analysis process 20 for image data can be used to identify the area of pressure-bearing regions in an image to obtain accurate quantitative data. The pressure analysis process 20 includes the following steps:
[0021] Step 202: Start;
[0022] Step 204: Input an image;
[0023] Step 206: Convert the image into a grayscale image;
[0024] Step 208: Group each pixel of the grayscale image to obtain a plurality of image data groups;
[0025] Step 210: Determine a compressed area of the image data according to the compressed state of the image;
[0026] Step 212: End.
[0027] The stress analysis system 10 of the embodiment of the present application can accurately identify and extract the stress area from the image according to the stress analysis process 20, thereby obtaining accurate quantitative data. For example, taking a microphone rubber as an example, after analysis by a machine learning model, the quantitative data of the area occupied by different stress ranges, such as Area_A and Area_B, are obtained. Figure 3 shown.
[0028] Depend on Figure 3 It can be seen that when the contact stress of the microphone rubber exceeds 1 MPa, the pressure areas of Area_A and Area_B are 0.095 mm 3 and 1.275mm 3 In this way, precise quantitative data can increase the accuracy of data analysis to highlight the significant differences in airtightness between products.
[0029] To reduce the complexity of clustering calculation, the image data compression analysis system 10 of the embodiment of the present application first converts the color image into a grayscale image, and then performs clustering using the grayscale value (ie, RGB value) of each pixel in the grayscale image to improve calculation efficiency.
[0030] In step 204 and step 206, the stress analysis system 10 inputs an image and converts the image into a grayscale image. In step 208, the stress analysis system 10 further groups each pixel of the grayscale image to obtain a plurality of image data groups.
[0031] In one embodiment, the present invention utilizes the K-means algorithm in the scikit-learn package for machine learning in the Python programming language for data analysis, such as the Lloyd and Elkan algorithms. In the example of the Lloyd algorithm, to ensure that data points (i.e., pixels in a grayscale image) within a group of the same image data set have similar characteristics while there are significant differences in characteristics between groups, it is necessary to find k sets of groups S_i in the data set (x_1, x_2, ..., x_n) such that the within-cluster sum of squares (WCSS) of S_i is minimized, as shown in Equation (1).
[0032]
[0033] where μ i It's S i The average of all points in .
[0034] First, a group of image data sets is initialized with center points. The k-means++ algorithm is used to ensure that the distances between the center points m_i of each data set are as far as possible. Therefore, the stress analysis system 10 of this embodiment randomly selects a first center point and calculates the distance between the first center point and other pixels. Then, using this distance as a parameter, an iterative method including probability is used to select the center points of other image data sets.
[0035] Next, each pixel (i.e., data point) in the image data group is divided into the group closest to the center point. It should be noted that each pixel in the image data group is assigned to only a single group. That is, the same pixel (i.e., data point) will not be assigned to multiple data groups at the same time.
[0036] After the above grouping step, the stress analysis system 10 of the embodiment of the present application recalculates the center point m of each data group using the following formula (2): i , to update the center points of all image data groups until the group center points no longer change, or the calculation is stopped after reaching the predetermined maximum number of iterations to determine the image data group that has been allocated.
[0037]
[0038] like Figure 4 As shown, Figure 4 This is a schematic diagram of grouping image data groups according to an embodiment of the present application.
[0039] The Lloyd algorithm clusters the image data, including setting K-means, clustering, and moving K-means. Convergence is then confirmed, and the image data clusters are clustered when they meet convergence criteria, i.e., a Voronoi diagram. In one embodiment, the convergence criteria can be that the center point of each image data cluster no longer changes, or that the number of clustering iterations meets a maximum number of iterations.
[0040] Because different initial center points may significantly affect the clustering results of the K-means algorithm, the stress analysis process 20 of the embodiment of the present application uses formula (1) to find a minimum intra-group sum of squares. In other words, the more times the clustering calculation is repeated, the more stable the clustering results can be. It is worth noting that because the stress analysis process 20 of the embodiment of the present application is based on grayscale images for clustering, the RGB values of the grayscale images can exhibit a one-dimensional linear distribution characteristic, thereby reducing the time required to execute the clustering algorithm.
[0041] Please refer to Figure 5 as well as Figure 6 , Figure 5This is a statistical diagram of the distribution of grayscale RGB values of an image before clustering using the K-means algorithm in an embodiment of the present application. Figure 6 This is a statistical diagram of the distribution of grayscale RGB values of an image after clustering using the K-means algorithm in an embodiment of the present application.
[0042] Depend on Figure 5 It can be seen that the grayscale RGB values of the image before clustering are evenly distributed with only one obvious peak, but the uniform color distribution cannot be used to calculate the compressed area of the image.
[0043] like Figure 6 As shown, after the RGB values of the grayscale image are divided into 12 groups by the K-means algorithm, the pressure analysis system 10 of the embodiment of the present application can perform statistics on the RGB values of different grayscale images and calculate the converted pressure area (i.e., step 210 of the pressure analysis process 20).
[0044] It is worth noting that the number of clusters used in the K-means algorithm is not limited to the above examples. Other numbers of groups can also be used in different scenarios without being limited to this.
[0045] In this way, the pressure analysis system 10 of the embodiment of the present application can perform image data analysis through machine learning, thereby eliminating errors caused by human estimation and ensuring the accuracy and consistency of the data.
[0046] It is worth noting that the above embodiments describe the concepts of the present application. Those skilled in the art may make appropriate modifications accordingly and are not limited to these. For example, other images that can distinguish data by color, such as pressure, object appearance defects, etc., grouping methods and the number of groups, etc. are not limited to the above examples and can be adjusted according to user needs and are all within the scope of the present application.
[0047] In summary, the embodiments of the present application provide a method for analyzing compression of image data and a related system thereof to quantify the compression analysis of image data. By utilizing a machine learning model to classify image pixels, the compression area of the image data is determined, thereby improving the efficiency of image data analysis.
[0048] The above description is only a preferred embodiment of the present application. All equivalent changes and modifications made within the scope of the patent application should fall within the scope of the claims of the present application.
Claims
1. A method for analyzing compression of image data, comprising: Input an image; converting the image into a grayscale image; Grouping each pixel of the grayscale image to obtain a plurality of image data groups; Determining a compression state of the image according to the plurality of image data groups; and A compressed area of the image is determined according to the compressed state of the image.
2. The method for analyzing image data under pressure according to claim 1, wherein: The step of grouping each grayscale of the grayscale image to obtain the plurality of image data groups includes: Randomly determining a first center point of the plurality of image data groups; Determining the center points of other image data groups of the plurality of image data groups according to the distances between the first center point and other pixels; Updating the center point of each of the plurality of image data groups; Determine whether the center point of each group converges; and When the center point of each group meets a convergence condition, the plurality of image data groups are determined.
3. The method for analyzing pressure of image data according to claim 2, wherein: The center point of each of the plurality of image data groups is updated in an iterative manner.
4. The method for analyzing image data under pressure according to claim 1, wherein: Each pixel belongs to one of the plurality of image data groups.
5. The method for analyzing pressure of image data according to claim 1, wherein: The convergence condition is that the center point of each group does not change, or a maximum number of iterations is satisfied, and the maximum number of iterations is a sum of squares of distances between each pixel of the grayscale image and the corresponding center point.
6. A compression analysis system for image data, comprising: a processing device; and A memory device is coupled to the processing device and is used to store a program code to instruct the processing device to execute a compression analysis process of an image, wherein The stress analysis process includes: Input an image; converting the image into a grayscale image; Grouping each pixel of the grayscale image to obtain a plurality of image data groups; Determining a compression state of the image according to the plurality of image data groups; and A compressed area of the image is determined according to the compressed state of the image.
7. The image data pressure analysis system according to claim 6, characterized in that: The step of grouping each grayscale of the grayscale image to obtain the plurality of image data groups includes: Randomly determining a first center point of the plurality of image data groups; Determining the center points of other image data groups of the plurality of image data groups according to the distances between the first center point and other pixels; Updating the center point of each of the plurality of image data groups; Determine whether the center point of each group converges; and When the center point of each group meets a convergence condition, the plurality of image data groups are determined.
8. The image data pressure analysis system according to claim 7, characterized in that: The center point of each of the plurality of image data groups is updated in an iterative manner.
9. The image data pressure analysis system according to claim 6, characterized in that: Each pixel belongs to one of the plurality of image data groups.
10. The image data pressure analysis system according to claim 6, wherein: The convergence condition is that the center point of each group does not change, or a maximum number of iterations is satisfied, and the maximum number of iterations is a sum of squares of distances between each pixel of the grayscale image and the corresponding center point.