Fuzzy clustering image segmentation method based on morphological reconstruction and membership filtering

Abstract

The invention discloses a fuzzy clustering image segmentation method based on morphological reconstruction and membership filtering, including the following steps: reading the original image f, performing fuzzy clustering on the gray level of the original image, randomly initializing the membership degree matrix U0, setting the cycle counter t=0, updating the clustering center vk, updating the membership degree matrix U (t+1), performing median filtering on the membership degree matrix U, and obtaining the image segmentation result P according to the membership degree matrix U '. The inventionis not limited by additional parameters except the size of the filtering window. The type of noise contained in the image does not need to be considered. There is no need to calculate the distance between the pixel and the cluster center in the neighborhood, so the computational complexity is low. The neighborhood space information is used effectively, and the image segmentation effect is good. Compared with the existing algorithm, the method of the invention has obvious advantages of segmentation effect and running time.

Description

technical field [0001] The invention relates to the field of image segmentation methods, in particular to a fuzzy clustering image segmentation method based on morphological reconstruction and membership filter. Background technique [0002] Brain MRI is a non-invasive and high-resolution imaging technique that has been widely used clinically. How to quickly and accurately segment various components in brain MRI images is particularly important. In recent decades, many new methods and ideas have been applied in the field of medical image segmentation, among which fuzzy C-means clustering method and superpixel segmentation algorithm are more representative. [0003] The superpixel segmentation algorithm uses spatial constraint information, has certain noise resistance, and can retain the original boundary information of the image, but the number of superpixel segmentation needs to be determined manually, which largely depends on empirical values ​​and will directly affect th...

AI Technical Summary

Problems solved by technology

[0004] Fuzzy C-means clustering algorithm (Fuzzy C-Mean clustering, FCM) was proposed by Bezdek (references: Bezdek J C, Ehrlich R, Full W, FCM: The fuzzy c-means clustering algorithm [J]. Computers & Geosciences, 1984.10 ( 2): 191-203.), compared with K-means clustering, it has stronger fuzziness and retains more original image information, and has a good segmentation effect on images without noise and offset fields, but it only Considering grayscale information does not consider spatial information, so images with complex textures and backgrounds or images damaged by noise cannot be segmented. In order to solve this problem, most algorithms try to introduce local spatial information into the FCM algorithm to improve segmentation accuracy. Ahmed et al. (Ahmed M N, Yamany S M, Mohamed N, et al., A modified fuzzy C-means algorithm for bias field estimation and segmentation of MRI data[J]. IEEE Transactions on Medical

Images