CN110942467A - Improved watershed image segmentation method based on PSO-FCM - Google Patents
Improved watershed image segmentation method based on PSO-FCM Download PDFInfo
- Publication number
- CN110942467A CN110942467A CN201910949690.8A CN201910949690A CN110942467A CN 110942467 A CN110942467 A CN 110942467A CN 201910949690 A CN201910949690 A CN 201910949690A CN 110942467 A CN110942467 A CN 110942467A
- Authority
- CN
- China
- Prior art keywords
- data
- pso
- fcm
- clustering
- carrying
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 18
- 238000003709 image segmentation Methods 0.000 title claims abstract description 16
- 201000010065 polycystic ovary syndrome Diseases 0.000 claims abstract description 7
- 238000001914 filtration Methods 0.000 claims abstract description 4
- 230000000877 morphologic effect Effects 0.000 claims abstract description 4
- 230000008030 elimination Effects 0.000 claims abstract description 3
- 238000003379 elimination reaction Methods 0.000 claims abstract description 3
- 230000011218 segmentation Effects 0.000 abstract description 14
- 230000000694 effects Effects 0.000 abstract description 9
- 238000003759 clinical diagnosis Methods 0.000 abstract description 3
- 238000003745 diagnosis Methods 0.000 abstract description 3
- 238000004364 calculation method Methods 0.000 description 10
- 238000005457 optimization Methods 0.000 description 6
- 239000002245 particle Substances 0.000 description 3
- 238000002474 experimental method Methods 0.000 description 2
- 238000005192 partition Methods 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 239000003550 marker Substances 0.000 description 1
- 238000002604 ultrasonography Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/40—Image enhancement or restoration using histogram techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/155—Segmentation; Edge detection involving morphological operators
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/194—Segmentation; Edge detection involving foreground-background segmentation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Probability & Statistics with Applications (AREA)
- Ultra Sonic Daignosis Equipment (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses an improved watershed image segmentation method based on PSO-FCM, which specifically comprises the following steps: 1. carrying out morphological filtering and denoising processing on the PCOS image data; 2. carrying out histogram equalization processing on the data subjected to noise elimination; 3. carrying out PSO-FCM clustering on the data after equalization processing; 4. and performing improved watershed image segmentation on the pictures after clustering. The invention can enhance the contrast of data to a great extent, so that the vesicle area is more prominent, thereby generating better segmentation effect and more clearly segmenting the edge of the vesicle. The whole process is automatically completed, the burden of a doctor can be greatly reduced clinically, and the doctor can further analyze the ultrasonic data of a patient on the basis of the image processed by the algorithm, so that the efficiency of clinical diagnosis is improved on one hand, and the accuracy of diagnosis can be increased on the other hand.
Description
Technical Field
The invention relates to a clustering optimized improved watershed image segmentation method, which is used in ultrasound image segmentation of polycystic ovary syndrome (PCOS). Firstly, histogram equalization is carried out on an ultrasonic image, image contrast is increased to a certain degree, then a PSO-FCM clustering method is used for carrying out clustering optimization on the ultrasonic image, and finally, an improved watershed is used for carrying out image segmentation operation on the ultrasonic image.
Background
Histogram equalization is an image enhancement method that increases the global contrast of an image, especially for grayscale images where the pixel values are relatively close, by which the luminance is better distributed over the histogram. Therefore, the histogram equalization can enhance the local contrast without influencing the overall contrast.
Compared with the common clustering algorithm, the fuzzy C-means clustering (FCM) algorithm is a flexible partition, wherein membership is introduced to represent the membership condition of a sample to each clustering center, and compared with the hard partition of the common clustering method, the FCM algorithm has a better clustering effect. The selection of the cluster center is the most important factor to influence the clustering effect.
The Particle Swarm Optimization (PSO) is an optimization algorithm, which starts from random solution, continuously iterates to search for an optimal solution, continuously searches for respective optimal solutions by different individuals, then counts the optimal solutions of the individuals to serve as global optimization, and searches for the optimal solution again through an updating formula until the iteration times are reached or the iteration ending condition is met, and outputs a final solution.
The watershed algorithm is a common algorithm in image segmentation, considers the segmentation of an image according to the composition of watersheds, and is a mathematical form segmentation algorithm based on a topological theory. The basic idea is to regard the image as a geodetic topological landform, the gray value of each pixel in the image represents the altitude of the point, each local minimum and the area of influence thereof are called catchment basins, and the boundaries of the catchment basins form watersheds. The traditional watershed algorithm based on the gradient image is easy to cause over-segmentation of the image, so that the improved watershed algorithm based on the marker is used for marking the foreground background of the image, and the over-segmentation phenomenon caused by noise is avoided.
Disclosure of Invention
The invention aims to realize automatic segmentation of a PCOS ultrasonic image, wherein the image is optimized before being input into a watershed and then a better segmentation effect is realized through an improved watershed algorithm. The invention is beneficial to improving the working efficiency of clinical diagnosis and improving the diagnosis accuracy to a certain extent.
According to the technical scheme provided by the invention, the improved watershed image segmentation method based on PSO-FCM optimization is provided, and comprises the following steps:
step 1, performing morphological filtering and denoising processing on PCOS image data;
step 2, carrying out histogram equalization processing on the data subjected to noise elimination;
step 3, carrying out PSO-FCM clustering on the data after the equalization processing;
step 4, performing improved watershed image segmentation on the clustered pictures;
in step 3, the step of performing PSO-FCM clustering on the image data after histogram equalization specifically includes:
(1) taking a pixel value as a clustering center, randomly initializing m populations, wherein each population has c clustering center individuals, and the pixel value of each individual is different;
(2) calculating the current membership value according to a calculation formula of the clustering center and the FCM membership;
(3) calculating and storing respective objective function values J1 of the m populations according to the cluster centers and the membership values obtained by the calculation in the step;
(4) according to the PSO speed and position updating formula, recalculating the clustering centers of the m populations, and calculating and storing corresponding objective function values J2 according to the step (2);
(5) comparing objective function values obtained by calculation in m populations, and for individuals in a single population, respectively selecting a cluster center corresponding to a smaller objective function value in J1 and J2 as pbest, and selecting a cluster center corresponding to a smallest objective function value J in the m populations as gbest;
(6) repeating the step (4) according to the obtained pbest and gbest until iteration reaches the iteration times or the objective function is smaller than a certain threshold value, ending the loop and keeping the best clustering center;
(7) clustering the data according to the reserved clustering centers;
the calculation formula for obtaining the membership degree according to the clustering center in the step 3(2) is as follows:
wherein xkDenotes the kth sample, eiRepresents the ith cluster center, m is a constant, and m is 2, muikRepresenting the membership value of the kth sample with respect to the ith cluster center.
Wherein the target function calculation formula of the FCM is as follows:
where n is the number of samples.
The velocity and position updating formula of the PSO in step 3(4) is as follows:
vj＝wvj+c1×rand()×(pbestj-zj)+c2×rand()×(gbestj-zj) (3)
zj＝zj+vj(4)
wherein formula (3) is a velocity update formula, formula (4) is a location update formula, c1,c2Is a constant, w is an inertia factor, is a constant, and rand () represents a random number between 0 and 1, pbestjRepresents the clustering center value of the J minimum in the jth population, gbest represents the clustering center value of the J minimum in all the populations, vjAnd zjRepresenting the velocity and position of individual particles.
In the step 4, the clustered pictures are subjected to improved watershed operation, and the method specifically comprises the following steps:
(1) firstly, carrying out binarization processing on the clustered data;
(2) performing expansion processing on the data after binarization processing to obtain a determined background area;
(3) calculating Euclidean distance between a non-zero point and a zero point in the binary image, setting a threshold value, and then performing threshold value processing according to the distance to obtain a determined foreground area;
(4) and (3) subtracting the foreground region determined in the step (3) from the background region determined in the step (2) to obtain an uncertain region of the vesicle, using the uncertain region as a mark, segmenting by using a watershed, and outputting a result.
Compared with the method for optimizing by not using clusters or only using FCM, the improved watershed image segmentation method using PSO-FCM optimization can enhance the contrast of data to a great extent, so that the vesicle area is more prominent, a better segmentation effect is achieved, and the edge of the vesicle is segmented more clearly. The whole process is automatically completed, the burden of a doctor can be greatly reduced clinically, and the doctor can further analyze the ultrasonic data of a patient on the basis of the image processed by the algorithm, so that the efficiency of clinical diagnosis is improved on one hand, and the accuracy of diagnosis can be increased on the other hand.
Drawings
FIG. 1 is a flow chart of the experiment.
FIGS. 2(a) - (d) are graphs of experimental watershed segmented image comparison results.
The specific implementation mode is as follows:
the present invention is further illustrated by the following specific examples. The following description is exemplary and explanatory only and is not restrictive of the invention in any way.
As shown in fig. 1, the steps of the present invention are as follows:
step 1, as image data used in an experiment is directly obtained from a hospital, and various noise signals exist in original data, firstly, morphological filtering and denoising processing is carried out on PCOS image data;
step 2, carrying out histogram equalization processing on the image data after the noise cancellation, and redistributing the pixel points which are concentrated in the histogram originally according to an accumulation function in the histogram to enhance the contrast;
step 3, performing PSO-FCM clustering on the data after equalization processing, wherein when clustering center searching is performed through PSO, in order to prevent the searched clustering center from crossing the border, a pixel value range of an original picture is obtained on the basis of the clustering center searching, and then the PSO optimized clustering center is limited in a normal range, so that border crossing is prevented, and meanwhile, the efficiency is improved;
step 4, performing improved watershed image segmentation on the clustered pictures;
in step 3, the step of performing PSO-FCM clustering on the picture data after histogram equalization specifically includes:
(1) taking a pixel value as a clustering center, randomly initializing m populations, wherein each population has c clustering center individuals, and the pixel value of each individual is different;
(2) calculating the current membership value according to a calculation formula of the clustering center and the FCM membership;
(3) calculating and storing respective objective function values J1 of the m populations according to the cluster centers and the membership values obtained by the calculation in the step;
(4) according to the PSO speed and position updating formula, recalculating the clustering centers of the m populations, and calculating and storing corresponding objective function values J2 according to the step (2);
(5) comparing objective function values obtained by calculation in m populations, and for individuals in a single population, respectively selecting a cluster center corresponding to a smaller objective function value in J1 and J2 as pbest, and selecting a cluster center corresponding to a smallest objective function value J in the m populations as gbest;
(6) repeating the step (4) according to the obtained pbest and gbest until iteration reaches the iteration times or the objective function is smaller than a certain threshold value, ending the loop and keeping the best clustering center;
(7) clustering the data according to the reserved clustering centers;
the calculation formula for obtaining the membership degree according to the clustering center in the step 3(2) is as follows:
wherein xkDenotes the kth sample, eiRepresents the ith cluster center, m is a constant, and m is 2, muikRepresenting the membership value of the kth sample with respect to the ith cluster center.
Wherein the target function calculation formula of the FCM is as follows:
where n is the number of samples.
The velocity and position updating formula of the PSO in step 3(4) is as follows:
vj＝wvj+c1×rand()×(Pbestj-zj)+c2×rand()×(gbestj-zj) (3)
zj＝j+vj(4)
wherein formula (3) is a velocity update formula, formula (4) is a location update formula, c1,c2Is a constant, w is an inertia factor, is a constant, and rand () represents a random number between 0 and 1, pbestjRepresents the clustering center value of the J minimum in the jth population, gbest represents the clustering center value of the J minimum in all the populations, vjAnd zjRepresenting the velocity and position of individual particles.
In the step 4, the clustered pictures are subjected to improved watershed operation, and the method specifically comprises the following steps:
(1) firstly, carrying out binarization processing on the clustered data;
(2) performing expansion processing on the data after binarization processing to obtain a determined background area;
(3) calculating Euclidean distance between a non-zero point and a zero point in the binary image, setting a threshold value, and then performing threshold value processing according to the distance to obtain a determined foreground area;
(4) and (3) subtracting the foreground region determined in the step (3) from the background region determined in the step (2) to obtain an uncertain region of the vesicle, using the uncertain region as a mark, segmenting by using a watershed, and outputting a result.
The images were subjected to different processes of FCM clustering and PSO-FCM clustering, and then to improved watershed, and the segmentation results are shown in FIG. 2. Wherein FIG. 2(a) shows the result of watershed segmentation, (b) shows the result of histogram equalization-watershed segmentation, (c) shows the result of histogram equalization-FCM-watershed segmentation, and (d) shows the result of histogram equalization-PSO-FCM-watershed segmentation.
By comparing the different methods described above, the PSO-FCM combined with the improved watershed algorithm has a clearer segmentation effect, and the effect is the best. Before the improved watershed is used for segmentation, histogram equalization, FCM clustering and PSO-FCM clustering are firstly carried out, so that the image contrast is obviously improved, and when the watershed is used finally, a good effect is also produced, the connection of segmentation lines among vesicles is reduced, the vesicles are independently segmented one by one, and a good auxiliary effect is achieved in clinical tests.
Claims (2)
1. The improved watershed image segmentation method based on the PSO-FCM is characterized by comprising the following steps of:
step 1, performing morphological filtering and denoising processing on PCOS image data;
step 2, carrying out histogram equalization processing on the data subjected to noise elimination;
step 3, carrying out PSO-FCM clustering on the data after the equalization processing;
and 4, performing improved watershed image segmentation on the clustered pictures.
2. The PSO-FCM-based improved watershed image segmentation method according to claim 1, wherein in step 4, the step of performing improved watershed image segmentation on the clustered data specifically comprises:
(1) firstly, carrying out binarization processing on the clustered data;
(2) performing expansion processing on the data after binarization processing to obtain a determined background area;
(3) calculating Euclidean distance between a non-zero point and a zero point in the binary image, setting a threshold value, and then performing threshold value processing according to the distance to obtain a determined foreground area;
(4) and (3) subtracting the foreground region determined in the step (3) from the background region determined in the step (2) to obtain an uncertain region of the vesicle, using the uncertain region as a mark, segmenting by using a watershed, and outputting a result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910949690.8A CN110942467A (en) | 2019-10-08 | 2019-10-08 | Improved watershed image segmentation method based on PSO-FCM |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910949690.8A CN110942467A (en) | 2019-10-08 | 2019-10-08 | Improved watershed image segmentation method based on PSO-FCM |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110942467A true CN110942467A (en) | 2020-03-31 |
Family
ID=69905916
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910949690.8A Pending CN110942467A (en) | 2019-10-08 | 2019-10-08 | Improved watershed image segmentation method based on PSO-FCM |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110942467A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112101461A (en) * | 2020-09-16 | 2020-12-18 | 北京邮电大学 | HRTF-PSO-FCM-based unmanned aerial vehicle reconnaissance visual information audibility method |
| CN112507991A (en) * | 2021-02-04 | 2021-03-16 | 季华实验室 | Method and system for setting gate of flow cytometer data, storage medium and electronic equipment |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102945553A (en) * | 2012-11-27 | 2013-02-27 | 西安电子科技大学 | Remote sensing image partition method based on automatic difference clustering algorithm |
| CN104651462A (en) * | 2015-01-29 | 2015-05-27 | 华南农业大学 | Method for detecting magnaporthe grisea spore based on microscopic image analysis |
-
2019
- 2019-10-08 CN CN201910949690.8A patent/CN110942467A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102945553A (en) * | 2012-11-27 | 2013-02-27 | 西安电子科技大学 | Remote sensing image partition method based on automatic difference clustering algorithm |
| CN104651462A (en) * | 2015-01-29 | 2015-05-27 | 华南农业大学 | Method for detecting magnaporthe grisea spore based on microscopic image analysis |
Non-Patent Citations (2)
| Title |
|---|
| P KAVITHA ET AL.: "A Novel Hybrid Segmentation Method with Particle Swarm Optimization and Fuzzy C-Mean Based On Partitioning the Image for Detecting Lung Cancer", 《INTERNATIONAL JOURNAL OF ENGINEERING AND ADVANCED TECHNOLOGY》 * |
| 李丹等: "基于FCM 和分水岭算法的无人机影像中林分因子提取", 《林业科学》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112101461A (en) * | 2020-09-16 | 2020-12-18 | 北京邮电大学 | HRTF-PSO-FCM-based unmanned aerial vehicle reconnaissance visual information audibility method |
| CN112101461B (en) * | 2020-09-16 | 2022-02-25 | 北京邮电大学 | HRTF-PSO-FCM-based unmanned aerial vehicle reconnaissance visual information audibility method |
| CN112507991A (en) * | 2021-02-04 | 2021-03-16 | 季华实验室 | Method and system for setting gate of flow cytometer data, storage medium and electronic equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Anjna et al. | Review of image segmentation technique | |
| CN111145209B (en) | Medical image segmentation method, device, equipment and storage medium | |
| CN108629783B (en) | Image segmentation method, system and medium based on image feature density peak search | |
| JP2006209755A (en) | Method for tracing moving object inside frame sequence acquired from scene | |
| CA2613116A1 (en) | Video object cut and paste | |
| Gupta et al. | Automatic trimap generation for image matting | |
| CN114492619B (en) | Point cloud data set construction method and device based on statistics and concave-convex performance | |
| CN110555868A (en) | method for detecting small moving target under complex ground background | |
| WO2019197021A1 (en) | Device and method for instance-level segmentation of an image | |
| Zheng et al. | Improvement of grayscale image 2D maximum entropy threshold segmentation method | |
| CN107194929B (en) | Method for tracking region of interest of lung CT image | |
| CN109636810B (en) | Pulmonary nodule segmentation method and system of CT image | |
| CN110942467A (en) | Improved watershed image segmentation method based on PSO-FCM | |
| Zhu et al. | Automatic object detection and segmentation from underwater images via saliency-based region merging | |
| CN111868783B (en) | Region merging image segmentation algorithm based on boundary extraction | |
| CN112435264A (en) | 42CrMo single-phase metallographic structure segmentation method and system based on deep learning | |
| CN107729863B (en) | Human finger vein recognition method | |
| Kachouie et al. | Watershed deconvolution for cell segmentation | |
| CN109241865B (en) | Vehicle detection segmentation algorithm under weak contrast traffic scene | |
| CN111612793A (en) | Automatic skull removing method for brain magnetic resonance image | |
| CN110599518B (en) | Target tracking method based on visual saliency and super-pixel segmentation and condition number blocking | |
| Dzyubachyk et al. | Model-based approach for tracking embryogenesis in Caenorhabditis elegans fluorescence microscopy data | |
| Ma et al. | Fully convolutional network with cluster for semantic segmentation | |
| CN109086774B (en) | Color image binarization method and system based on naive Bayes | |
| CN111242960A (en) | Image segmentation method based on complex network theory |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |