An Adaptive Kernel Clustering Image Segmentation Method

Abstract

The invention discloses a self-adaptive kernel cluster image partitioning method which aims at combining particle swarms, fuzzy clustering and granularity validity functions and providing a monitoring-free gray image partitioning method. The self-adaptive kernel cluster image partitioning method comprises the following steps: according to the characteristic difference of different image areas, corresponding areas which have identical characteristics or similarity within a certain area in images into one type of clustering processing through clustering, thereby completing the image partitioning process. Chaos particles are adopted as an initial center of clustering, so that the sensitivity of the partitioning result to the selection of the initial clustering center is alleviated, and the possibility that an algorithm falls into partial convergence is reduced. In the fuzzy clustering process, the gray level of the images instead of that of image pixel points is taken as a clustering sample set, and the image partitioning speed is increased; in the clustering process, the validity functions of clustering is dynamically calculated, an optimal clustering number is selected, and the subjectivity in artificially appointing the clustering number is avoided.

Description

technical field [0001] The invention relates to an adaptive kernel clustering image segmentation method, which belongs to the technical field of image processing. Background technique [0002] Image segmentation is to extract the target area of ​​interest in the image based on the feature difference between the image areas, so that the subsequent tasks can be carried out, and it is an important basis for image processing. In practical applications, due to the equipment, environment, system and other factors of image acquisition, images inevitably have fuzziness and uncertainty, and fuzzy theory is very suitable for describing such characteristics of images. Clustering is to divide a collection of objects without specific categories into several small sub-collections according to certain requirements or rules, and requires that the differences between the sub-collections be as large as possible. Introducing the combination of fuzzy theory and clustering theory into the image...

AI Technical Summary

Problems solved by technology

Using the peaks and troughs in the gray histogram of the image to be segmented to determine the number of clusters of the image to be segmented can ensure that the pixels in each class have similar gray values, but when the boundary between the peak and the trough is blurred , the number of clusters obtained by this method will be inaccurate, resulting in wrong segmentation results

Images