An intelligent image segmentation method based on fuzzy C-means clustering
By optimizing the FCM algorithm and information entropy correction membership, the problems of traditional fuzzy C-mean clustering being sensitive to noise and insufficient Euclidean distance are solved, and a more accurate image segmentation effect is achieved.
Patent Information
- Application Number
- CN202411376488.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-30
- Publication Date
- 2024-12-06
- Estimated Expiration
- 2044-09-30
AI Technical Summary
The traditional fuzzy C-mean clustering algorithm is sensitive to noise and outliers, affecting the accuracy of clustering results, and judging membership based on Euclidean distance alone, and it is impossible to effectively distinguish between bone and muscle tissue.
By optimizing the objective function of the FCM algorithm, combining the grayscale value difference of pixel points and Euclidean distance, the clustered blur pixel points in the preliminary clustering results are obtained, and the membership parameters are constructed using information entropy and Euclidean distance. To divide the bone and muscle tissue areas, combine information entropy for image enhancement and segmentation.
It improves the accuracy and segmentation effect of clustering, reduces the negative impact of noise and outliers on the results, and obtains more accurate image segmentation results.
Smart Images

Figure CN118887407B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of image data processing, and in particular to an image intelligent segmentation method based on fuzzy C-means clustering. Background Art
[0002] Traditional clustering algorithms are often hard clustering methods, where each data point can only belong to one cluster. Fuzzy clustering algorithms are clustering analysis methods based on fuzzy mathematics, which allow each data point to have a corresponding probability of being assigned to different clusters, rather than just being assigned to a certain cluster. This flexibility makes fuzzy clustering more applicable than traditional hard clustering in some cases.
[0003] Among various fuzzy clustering algorithms, the fuzzy C-means clustering algorithm (hereinafter referred to as the FCM algorithm) is the most widely used. Its basic idea is to determine the membership of the cluster center and each data point by minimizing the weighted square error between the data point and the cluster center. When this algorithm is used for image segmentation, it is a labeling process after unsupervised fuzzy clustering, which can reduce human intervention and is very suitable for medical image segmentation scenarios. The FCM algorithm is sensitive to noise and outliers, and these data points may affect the accuracy of the final clustering. Summary of the invention
[0004] The invention provides an image intelligent segmentation method based on fuzzy C-means clustering to solve the existing problems.
[0005] The present invention provides an image intelligent segmentation method based on fuzzy C-means clustering using the following technical solutions:
[0006] An embodiment of the present invention provides an image intelligent segmentation method based on fuzzy C-means clustering, the method comprising the following steps:
[0007] Obtaining a hand bone grayscale image, and determining an initial cluster center and the number of clusters according to edge information in the hand bone grayscale image;
[0008] According to the gray value difference between the pixel points in the hand bone grayscale image and the initial clustering center, the objective function of the FCM algorithm is optimized, and the pixel points in the hand bone grayscale image are clustered to obtain a preliminary clustering result, and the clustered fuzzy pixel points in the preliminary clustering result are obtained;
[0009] Using the Euclidean distance between the clustering fuzzy pixel point and the cluster center of the cluster cluster in the preliminary clustering result and the gray value of the pixel point in the hand bone grayscale image, the fuzzy membership of the clustering fuzzy pixel point in the preliminary clustering result is corrected to obtain the final clustering result;
[0010] The image grayscale is enhanced and image segmentation is performed based on the information entropy of the clusters in the final clustering results.
[0011] Furthermore, according to the gray value difference between the pixel points in the hand bone grayscale image and the initial clustering center, the objective function of the FCM algorithm is optimized, and the pixel points in the hand bone grayscale image are clustered to obtain a preliminary clustering result, and the clustered fuzzy pixel points in the preliminary clustering result are obtained, including the specific method of:
[0012]
[0013] in, represents the optimized objective function, represents the optimized membership matrix, represents the initial cluster center, Indicates the preset number of clusters, , represents the preset fuzzy parameters, Indicates The pixels belong to The fuzzy membership of the initial cluster centers, Indicates Pixels to The Euclidean distance of the initial cluster centers, Indicates Pixels and The gray value difference of the initial cluster centers, is the number of pixels in the hand bone grayscale image;
[0014] Based on the FCM algorithm, according to the optimized membership degree and the calculation formula of the initial cluster center, the pixels in the hand bone grayscale image are clustered through the iterative updating method using the objective function optimized by the FCM algorithm to obtain the preliminary clustering results. The pixels in the preliminary clustering results that belong to several clusters at the same time are recorded as clustering fuzzy pixels.
[0015] Furthermore, the calculation formula of the optimized membership degree and initial cluster center includes the specific calculation method as follows:
[0016]
[0017]
[0018] in, Indicates The pixels belong to The fuzzy membership of the initial cluster centers, Indicates Pixels to The Euclidean distance of the initial cluster centers, Indicates Pixels to The Euclidean distance of the initial cluster centers, Indicates Pixels and The gray value difference of the initial cluster centers, Indicates Pixels and The gray value difference of the initial cluster centers;
[0019] Indicates Initial cluster centers, Indicates The gray value of a pixel, Indicates The pixels belong to The fuzzy membership of the initial cluster centers, Indicates the number of pixels.
[0020] Furthermore, the fuzzy membership of the clustering fuzzy pixels in the preliminary clustering results is corrected by using the Euclidean distance between the clustering fuzzy pixels and the clustering centers of the clusters in the preliminary clustering results and the grayscale values of the pixels in the hand bone grayscale image to obtain the final clustering results, including the following specific methods:
[0021] First, in the preliminary clustering results, the clusters were determined to belong to the bone area or muscle tissue area based on information entropy;
[0022] Secondly, the Euclidean distance between the clustered fuzzy pixel and the nearest skeleton line is calculated; the grayscale gradient of the hand bone grayscale image is obtained using the Sobel operator to construct The clustered fuzzy pixel point is located at the center of the rectangular window, and the grayscale complexity within the rectangular window is calculated, where is the preset first parameter;
[0023] Then, the membership parameters of clustered fuzzy pixels are constructed according to the Euclidean distance between the clustered fuzzy pixels and the nearest skeleton line and the grayscale complexity of the rectangular window where the clustered fuzzy pixels are located.
[0024] Finally, according to the size of the membership parameters of the clustered fuzzy pixels, the clustered fuzzy pixels are divided into pixels belonging to the bone area and pixels belonging to the muscle tissue area.
[0025] Furthermore, the specific method of determining whether the cluster belongs to the bone region or the muscle tissue region according to the information entropy is as follows:
[0026] Calculate the information entropy of any cluster and set the threshold , the information entropy is greater than or equal to the threshold The clusters with information entropy less than the threshold are recorded as skeleton regions. The clusters are recorded as muscle tissue areas.
[0027] Furthermore, the grayscale complexity within the rectangular window is calculated, including the following specific methods:
[0028]
[0029] in, Indicates the The grayscale complexity within the rectangular window centered on the clustered fuzzy pixel point; Indicates the The mean gray gradient in the window centered on the clustered fuzzy pixel; Indicates the The pixels in the window centered on the clustered fuzzy pixel and the central pixel The gray value difference.
[0030] Furthermore, the construction of membership parameters of clustered fuzzy pixels includes the following specific methods:
[0031]
[0032] in For the The membership parameters of clustered fuzzy pixels, Indicates The Euclidean distance between the clustered fuzzy pixel and the nearest skeleton line, Indicates the The grayscale complexity within the rectangular window centered at the clustered fuzzy pixel point, Represents the normalization function.
[0033] Furthermore, the clustering fuzzy pixels are divided into pixels belonging to the bone area and pixels belonging to the muscle tissue area according to the size of the membership parameter of the clustering fuzzy pixels, and the specific method includes:
[0034] Setting the Threshold If the membership parameter value of the clustered fuzzy pixel point is greater than the threshold value T2, it is determined that the clustered fuzzy pixel point is a pixel point belonging to the bone area. If the membership parameter value of the clustered fuzzy pixel point is less than or equal to the threshold value T2, it is determined that the clustered fuzzy pixel point is a pixel point belonging to the muscle tissue area.
[0035] Furthermore, the image grayscale is enhanced by combining the information entropy of the clusters in the final clustering result, and the image is segmented, including the following specific methods:
[0036] The image grayscale is linearly enhanced according to the information entropy of the pixel points belonging to the cluster and the grayscale value of the pixel points, so as to obtain an enhanced hand bone grayscale image; the information entropy of the pixel points belonging to the cluster is positively correlated with the grayscale value of the pixel points in the enhanced hand bone grayscale image;
[0037] Then, the enhanced hand bone grayscale image is segmented using the Otsu method to obtain accurate segmentation results.
[0038] Furthermore, the method of obtaining the hand bone grayscale image and determining the initial cluster center and the number of clusters according to the edge information in the hand bone grayscale image includes the following specific methods:
[0039] The patient's palm is placed flat, and the hand is photographed from above to obtain the original X-ray image of the patient's hand, which is recorded as the hand bone grayscale image. Then, the skeleton line of the hand bone grayscale image is extracted using the maximum inscribed circle method, and the midpoint of each skeleton line in the image is used as the initial cluster center. The number of clusters is determined based on the initial cluster centers.
[0040] The beneficial effect of the technical solution of the present invention is: compared with the traditional fuzzy C-means clustering algorithm that only judges the membership based on the Euclidean distance between the pixel point and the initial cluster center, the embodiment of the present invention improves the FCM algorithm in combination with the grayscale value of the pixel point to obtain a better clustering effect, and further for the clustered fuzzy pixel points in the preliminary clustering results, by constructing membership parameters, the membership of the clustered fuzzy pixel points in the preliminary clustering results is corrected to reduce the negative impact of these data on the clustering results, obtain the final clustering results, and then obtain a more accurate image segmentation effect. BRIEF DESCRIPTION OF THE DRAWINGS
[0041] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0042] Figure 1 The present invention is a flowchart of the steps of an image intelligent segmentation method based on fuzzy C-means clustering. DETAILED DESCRIPTION
[0043] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the specific implementation method, structure, features and effects of an image intelligent segmentation method based on fuzzy C-means clustering proposed by the present invention in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.
[0044] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
[0045] The specific scheme of the image intelligent segmentation method based on fuzzy C-means clustering provided by the present invention is described in detail below with reference to the accompanying drawings.
[0046] See also Figure 1 , which shows a flowchart of a method for intelligent image segmentation based on fuzzy C-means clustering provided by an embodiment of the present invention, the method comprising the following steps:
[0047] Step S001: Obtain a hand bone grayscale image, and determine the initial cluster center and the number of clusters according to edge information in the hand bone grayscale image.
[0048] It should be noted that in the traditional FCM algorithm, the selection of the initial cluster center is often random, which is in line with the idea of unsupervised learning, that is, no human intervention is required. However, random selection may always select an initial point that deviates greatly from the expected result, which will cause too many iterations or unsatisfactory final clustering results, etc. In order to enable those skilled in the art to better understand the present method, a hand X-ray is selected as an example to specifically describe the present invention.
[0049] Specifically, in order to implement the image intelligent segmentation method based on fuzzy C-means clustering proposed in this embodiment, it is first necessary to collect image information of the X-ray film. The specific process is as follows:
[0050] First, the patient's palm is placed flat, and the hand is photographed from above to obtain the original X-ray image of the patient's hand, which is recorded as the hand bone grayscale image; then, the hand bone grayscale image is subjected to skeleton line extraction using the maximum inscribed circle method, and the midpoint of each skeleton line in the image is used as the initial cluster center, and the number of clusters is determined based on the initial cluster centers. In this embodiment, a total of six skeleton lines are obtained, and six initial cluster centers are obtained.
[0051] The specific method of skeleton line extraction is: use Canny algorithm to perform edge detection on the hand bone grayscale image to obtain a number of edge lines, record the area formed by any closed edge line as the hand region, select circles that are completely contained in the hand region and are tangent to the hand region boundary at least at two points, the continuous points formed by the centers of these circles are the skeleton lines, and obtain the midpoint of the skeleton lines.
[0052] It should be noted that in the hand bone grayscale image, the five metacarpal bones of the hand are clearly distinguished, and the muscle tissue is distributed around the bones, so the skeleton line of the hand bone grayscale image can well represent the morphological characteristics of the hand when the skeleton is extracted.
[0053] So far, the initial cluster center and cluster number of the hand bone grayscale image are obtained through the above method.
[0054] Step S002: According to the gray value difference between the pixel points in the hand bone grayscale image and the initial clustering center, the objective function of the FCM algorithm is optimized, and the pixel points in the hand bone grayscale image are clustered to obtain a preliminary clustering result, and the clustered fuzzy pixel points in the preliminary clustering result are obtained.
[0055] It should be noted that in the FCM algorithm, the objective function and the constraints of the objective function are:
[0056]
[0057]
[0058] in, represents the membership matrix, represents the initial cluster center, represents the fuzzy parameter, , Indicates the number of categories divided, , Indicates The pixels belong to The fuzzy membership of the cluster centers is Indicates Pixels to The fuzzy parameter determines the fuzziness of the clustering, that is, the degree to which a data point can be divided into multiple clusters. The result is optimal when the minimum value is obtained. is the number of pixels in the hand bone grayscale image.
[0059] According to the objective function of the FCM algorithm mentioned above, the objective function of the traditional FCM algorithm only determines the degree of membership based on the Euclidean distance between the pixel point and the initial cluster center. The embodiment of the present invention improves the FCM algorithm in combination with the gray value of the pixel point to obtain a better clustering effect.
[0060] It should be noted that in the objective function of the FCM algorithm, the Euclidean distance between the pixel and the cluster center is used as the clustering rule. However, in the hand X-ray image, the segmentation target is the bone and muscle tissue, and the muscle tissue surrounds the bone. Therefore, using only the Euclidean distance as the clustering criterion obviously cannot meet the requirements of X-ray segmentation, so the objective function needs to be optimized.
[0061] Specifically, as an embodiment, the optimized objective function is as follows:
[0062]
[0063] in, represents the optimized objective function, represents the optimized membership matrix, represents the initial cluster center, Indicates the preset number of clusters, , represents the preset fuzzy parameters, Indicates The pixels belong to The fuzzy membership of the initial cluster centers, Indicates Pixels to The Euclidean distance of the initial cluster centers, Indicates Pixels and The gray value difference of the initial cluster centers, is the number of pixels in the hand bone grayscale image.
[0064] It should be noted that the fuzzy parameters are preset based on experience in this embodiment. , which can be adjusted according to actual conditions and is not specifically limited in this embodiment.
[0065] It should be noted that the pixels in the bone area not only show the characteristics of clustering in space, but also show great similarity in grayscale, that is, the pixels in the bone area are all located in the high grayscale interval, and the difference in grayscale values is small, while the muscle tissue is all located in the low grayscale interval, and the difference in grayscale values is large.
[0066] As an embodiment, the calculation formula of the optimized membership degree and the initial cluster center is:
[0067]
[0068]
[0069] in, Indicates The pixels belong to The fuzzy membership of the initial cluster centers, Indicates Pixels to The Euclidean distance of the initial cluster centers, Indicates Pixels to The Euclidean distance of the initial cluster centers, Indicates Pixels and The gray value difference of the initial cluster centers, Indicates Pixels and The gray value difference of the initial cluster centers.
[0070] Indicates Initial cluster centers, Indicates The gray value of a pixel, Indicates The pixels belong to The fuzzy membership of the initial cluster centers, Indicates the number of pixels.
[0071] Based on the FCM algorithm, according to the optimized membership degree and the calculation formula of the initial cluster center, the pixels in the hand bone grayscale image are clustered through the iterative updating method using the objective function optimized by the FCM algorithm to obtain the preliminary clustering results. The pixels in the preliminary clustering results that belong to several clusters at the same time are recorded as clustering fuzzy pixels.
[0072] The iterative update stop conditions include: the number of iterative updates reaches a preset number threshold, or .
[0073] in, Indicates The value of the objective function after the iteration update, Indicates The value of the objective function after the iteration update, is a preset second parameter, indicating an error threshold.
[0074] Step S003: using the Euclidean distance between the clustering fuzzy pixel point and the cluster center of the cluster cluster in the preliminary clustering result and the grayscale value of the pixel point in the hand bone grayscale image, the fuzzy membership of the clustering fuzzy pixel point in the preliminary clustering result is corrected to obtain the final clustering result.
[0075] It should be noted that, when observing the grayscale image of the hand bone, the overall grayscale value of the periosteum bone part in the bone area is relatively large, while the grayscale value of the bone marrow part is slightly different from that of the muscle tissue. Therefore, in the preliminary clustering results, the membership of the pixels located in the bone marrow part to two clusters will be very similar, which will cause the pixels corresponding to the bone cluster to be incorrectly divided into the cluster corresponding to the muscle tissue. Therefore, further feature analysis of pixels with similar membership is required.
[0076] It should be noted that there is a clear difference in the spatial distribution of the bone marrow pixels and the muscle tissue pixels. The bone marrow pixels are concentrated in the middle of the bone area, extending along the skeleton line and surrounded by the periosteum and bone area, while the muscle tissue pixels are more evenly distributed around the bone area. The Euclidean distance between the bone marrow pixels and the corresponding skeleton line is shorter than that of the muscle area. At the same time, the grayscale distribution of the bone marrow pixels in their neighborhood is more chaotic than that of the muscle tissue pixels.
[0077] First, in the preliminary clustering results, it is determined whether the cluster belongs to the bone area or the muscle tissue area based on the information entropy.
[0078] As an embodiment, the information entropy of any cluster is calculated as follows:
[0079]
[0080] in, Indicates The information entropy of clusters, Indicates The first The probability that the gray value of a pixel is in the cluster, Represents the base 2 logarithmic function.
[0081] Setting the Threshold , the information entropy is greater than or equal to the threshold The clusters with information entropy less than the threshold are recorded as skeleton regions. The clusters are recorded as muscle tissue areas.
[0082] It should be noted that the threshold value for selecting cluster information entropy is preset based on experience in this embodiment. , which can be adjusted according to actual conditions and is not specifically limited in this embodiment.
[0083] It should be noted that the smaller the probability of the grayscale value of a pixel in the entire cluster, the more uniform the grayscale distribution of the pixel, and the greater the information entropy of the cluster; the greater the probability of the grayscale value of a pixel in the entire cluster, the more chaotic the grayscale distribution of the pixel, and the smaller the information entropy of the cluster.
[0084] Secondly, the Euclidean distance between the clustered fuzzy pixel and the nearest skeleton line is calculated; the grayscale gradient of the hand bone grayscale image is obtained using the Sobel operator to construct The clustered fuzzy pixel point is located at the center of the rectangular window, and the grayscale complexity within the rectangular window is calculated, where is the first parameter of the preset.
[0085] It should be noted that this embodiment is preset based on experience , that is, the window size is , which can be adjusted according to actual conditions and is not specifically limited in this embodiment.
[0086] It should be noted that for the pixels of the bone marrow in the image, since the characteristics of the bone marrow area and the muscle tissue area are similar, they may be classified into both clusters at the same time. The pixels of the bone marrow will have two memberships with similar results. and The clustering result of the pixel point cannot be clearly judged only by the membership degree. The pixel point is a controversial point and needs further analysis.
[0087] As an embodiment, a specific method for calculating the grayscale complexity within a rectangular window centered on the clustered fuzzy pixel point is as follows:
[0088]
[0089]
[0090]
[0091] in, Indicates the The grayscale complexity within the rectangular window centered on the clustered fuzzy pixel point; Indicates the The mean gray gradient in the window centered on the clustered fuzzy pixel; Indicates the The pixels in the window centered on the clustered fuzzy pixel and the central pixel The gray value difference is Indicates The gray value of clustered fuzzy pixels, Indicates the The clustered fuzzy pixel point is the center of the window. The gray value of a pixel, represents the absolute value symbol, Represents the grayscale gradient mean.
[0092] It should be noted that the smaller the Euclidean distance between the clustered fuzzy pixel and the skeleton line, the greater the possibility that the clustered fuzzy pixel belongs to the bone marrow area; the greater the grayscale complexity of the window where the clustered fuzzy pixel is located, the greater the possibility that the clustered fuzzy pixel belongs to the bone marrow area.
[0093] Then, the membership parameters of the clustered fuzzy pixels are constructed according to the Euclidean distance between the clustered fuzzy pixels and the nearest skeleton line and the grayscale complexity of the rectangular window where the clustered fuzzy pixels are located.
[0094] Specifically, as an embodiment, the method for calculating the membership parameter of clustered fuzzy pixels is:
[0095]
[0096] in For the The membership parameters of clustered fuzzy pixels, Indicates The Euclidean distance between the clustered fuzzy pixel and the nearest skeleton line, Indicates the The grayscale complexity within the rectangular window centered at the clustered fuzzy pixel point, Indicates that the calculation results are normalized.
[0097] Finally, according to the size of the membership parameters of the clustered fuzzy pixels, the clustered fuzzy pixels are divided into pixels belonging to the bone area and pixels belonging to the muscle tissue area.
[0098] Specifically, set the threshold , filter out the pixels with fuzzy clustering, if the membership parameter value is greater than the threshold , determine that the clustered fuzzy pixel is a pixel belonging to the bone area, if the membership parameter value is less than or equal to the threshold , determine that the clustered fuzzy pixel points are pixel points belonging to the muscle tissue area.
[0099] It should be noted that the present embodiment sets the threshold of the membership parameter of the cluster fuzzy pixel point to which the pixel point belongs based on experience. , which can be adjusted according to actual conditions and is not specifically limited in this embodiment.
[0100] So far, the image segmentation of the hand X-ray film has been completed based on the fuzzy C-means clustering through the above method, and the bone marrow pixels and muscle tissue pixels that are difficult to distinguish in the image have been corrected to obtain accurate clustering results.
[0101] Step S004: performing image enhancement on the image grayscale and image segmentation based on the information entropy of the clusters in the final clustering result.
[0102] It should be noted that when making medical diagnoses, doctors often pay the most attention to the bone part, so the bone part needs to be enhanced in the image to highlight the contrast between the bone area and the muscle tissue that is not of interest.
[0103] The image grayscale is linearly enhanced according to the information entropy of the pixel points belonging to the cluster and the grayscale value of the pixel points to obtain an enhanced hand bone grayscale image; the information entropy of the pixel points belonging to the cluster is positively correlated with the grayscale value of the pixel points in the enhanced hand bone grayscale image.
[0104] Specifically, as an embodiment, the specific method for grayscale enhancement of the pixel points in the hand bone grayscale image is:
[0105]
[0106] in, Indicates The first The gray value of the pixel after correction. Indicates The information entropy of clusters, Indicates Clusters The gray value of a pixel.
[0107] It should be noted that, by utilizing the possibility that the cluster where the pixel point is located belongs to the bone area, the gray value of the pixel point is corrected to increase the gray value difference between the bone area and the muscle tissue area.
[0108] Then, the enhanced hand bone grayscale image is segmented using the Otsu method to obtain accurate segmentation results.
[0109] At this point, the embodiment of the present invention is completed.
[0110] Compared with the traditional fuzzy C-means clustering algorithm which only judges the membership based on the Euclidean distance between the pixel point and the initial cluster center, the embodiment of the present invention improves the FCM algorithm in combination with the grayscale value of the pixel point to obtain a better clustering effect, and further corrects the membership of the clustered fuzzy pixel points in the preliminary clustering result by constructing membership parameters to reduce the negative impact of these data on the clustering result, thereby obtaining the final clustering result and obtaining a more accurate image segmentation effect.
[0111] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.
Claims
1. A method for intelligent image segmentation based on fuzzy C-means clustering, characterized in that: The method comprises the following steps: Obtaining a hand bone grayscale image, and determining an initial cluster center and the number of clusters according to edge information in the hand bone grayscale image; According to the gray value difference between the pixel points in the hand bone grayscale image and the initial clustering center, the objective function of the FCM algorithm is optimized, and the pixel points in the hand bone grayscale image are clustered to obtain a preliminary clustering result, and the clustered fuzzy pixel points in the preliminary clustering result are obtained; Using the Euclidean distance between the clustering fuzzy pixel point and the cluster center of the cluster cluster in the preliminary clustering result and the gray value of the pixel point in the hand bone grayscale image, the fuzzy membership of the clustering fuzzy pixel point in the preliminary clustering result is corrected to obtain the final clustering result; Combine the information entropy of the clusters in the final clustering result to enhance the image grayscale and perform image segmentation; The objective function of the FCM algorithm is optimized according to the gray value difference between the pixel points in the hand bone grayscale image and the initial clustering center, and the pixel points in the hand bone grayscale image are clustered to obtain a preliminary clustering result, and the clustered fuzzy pixel points in the preliminary clustering result are obtained, including the specific method of: in, represents the optimized objective function, represents the optimized membership matrix, represents the initial cluster center, Indicates the preset number of clusters, , represents the preset fuzzy parameters, Indicates The pixels belong to The fuzzy membership of the initial cluster centers, Indicates Pixels to The Euclidean distance of the initial cluster centers, Indicates Pixels and The gray value difference of the initial cluster centers, is the number of pixels in the hand bone grayscale image; Based on the FCM algorithm, according to the optimized membership degree and the calculation formula of the initial cluster center, through the iterative update method, the pixel points in the hand bone grayscale image are clustered using the objective function optimized by the FCM algorithm to obtain the preliminary clustering results. The pixel points in the preliminary clustering results that belong to several clusters at the same time are recorded as cluster fuzzy pixels. The method of using the Euclidean distance between the clustering fuzzy pixel point and the cluster center of the cluster cluster in the preliminary clustering result and the gray value of the pixel point in the hand bone grayscale image to correct the fuzzy membership of the clustering fuzzy pixel point in the preliminary clustering result to obtain the final clustering result includes the following specific methods: First, in the preliminary clustering results, the clusters were determined to belong to the bone area or muscle tissue area based on information entropy; Secondly, the Euclidean distance between the clustered fuzzy pixel and the nearest skeleton line is calculated; the grayscale gradient of the hand bone grayscale image is obtained using the Sobel operator to construct The clustered fuzzy pixel point is located at the center of the rectangular window, and the grayscale complexity within the rectangular window is calculated, where is the preset first parameter; Then, the membership parameters of clustered fuzzy pixels are constructed according to the Euclidean distance between the clustered fuzzy pixels and the nearest skeleton line and the grayscale complexity of the rectangular window where the clustered fuzzy pixels are located. Finally, according to the size of the membership parameter of the clustered fuzzy pixels, the clustered fuzzy pixels are divided into pixels belonging to the bone area and pixels belonging to the muscle tissue area; The specific method of calculating the grayscale complexity within the rectangular window is as follows: in, Indicates the The grayscale complexity within the rectangular window centered on the clustered fuzzy pixel point; Indicates the The mean gray gradient in the window centered on the clustered fuzzy pixel; Indicates the The pixels in the window centered on the clustered fuzzy pixel and the central pixel The gray value difference of The specific method of constructing the membership parameters of clustered fuzzy pixels includes: in For the The membership parameters of clustered fuzzy pixels, Indicates The Euclidean distance between the clustered fuzzy pixel and the nearest skeleton line, Represents the normalization function.
2. According to claim 1, a method for intelligent image segmentation based on fuzzy C-means clustering, characterized in that: The calculation formula of the optimized membership degree and initial cluster center includes the specific calculation method as follows: in, Indicates The pixels belong to The fuzzy membership of the initial cluster centers, Indicates Pixels to The Euclidean distance of the initial cluster centers, Indicates Pixels to The Euclidean distance of the initial cluster centers, Indicates Pixels and The gray value difference of the initial cluster centers, Indicates Pixels and The gray value difference of the initial cluster centers; Indicates Initial cluster centers, Indicates The gray value of a pixel, Indicates The pixels belong to The fuzzy membership of the initial cluster centers, Indicates the number of pixels.
3. The image intelligent segmentation method based on fuzzy C-means clustering according to claim 1, characterized in that: The specific method of determining that the cluster belongs to the bone region or the muscle tissue region according to the information entropy is as follows: Calculate the information entropy of any cluster and set the threshold , the information entropy is greater than or equal to the threshold The clusters with information entropy less than the threshold are recorded as skeleton regions. The clusters of are recorded as muscle tissue areas.
4. The image intelligent segmentation method based on fuzzy C-means clustering according to claim 1, characterized in that: The method of dividing the clustered fuzzy pixels into pixels belonging to the bone area and pixels belonging to the muscle tissue area according to the size of the membership parameter of the clustered fuzzy pixels includes: Setting the Threshold If the membership parameter value of the clustered fuzzy pixel point is greater than the threshold value T2, it is determined that the clustered fuzzy pixel point is a pixel point belonging to the bone area. If the membership parameter value of the clustered fuzzy pixel point is less than or equal to the threshold value T2, it is determined that the clustered fuzzy pixel point is a pixel point belonging to the muscle tissue area.
5. The method for intelligent image segmentation based on fuzzy C-means clustering according to claim 1, characterized in that: The method of combining the information entropy of the clusters in the final clustering result to enhance the image grayscale and perform image segmentation includes the following specific methods: The image grayscale is linearly enhanced according to the information entropy of the pixel points belonging to the cluster and the grayscale value of the pixel points, so as to obtain an enhanced hand bone grayscale image; the information entropy of the pixel points belonging to the cluster is positively correlated with the grayscale value of the pixel points in the enhanced hand bone grayscale image; Then, the enhanced hand bone grayscale image is segmented using the Otsu method to obtain accurate segmentation results.
6. The method for intelligent image segmentation based on fuzzy C-means clustering according to claim 1, characterized in that: The method of obtaining the hand bone grayscale image and determining the initial cluster center and the number of clusters according to the edge information in the hand bone grayscale image includes: The patient's palm is placed flat, and the hand is photographed from above to obtain the original X-ray image of the patient's hand, which is recorded as the hand bone grayscale image. Then, the skeleton line of the hand bone grayscale image is extracted using the maximum inscribed circle method, and the midpoint of each skeleton line in the image is used as the initial cluster center. The number of clusters is determined based on the initial cluster centers.
