Image segmentation method based on quaternion and fuzzy C-means clustering

Abstract

The invention discloses an image segmentation method based on quaternion and fuzzy C-means clustering. The method includes the steps that an image to be segmented is converted into quaternion space, clustering amount is specified, a membership matrix U is initialized, the different degree between a current clustering center and pixels is measured through the defined quaternion distance, an objective function, a novel membership matrix and a clustering center are calculated, if the objective function is smaller than or equal to an iteration stopping threshold value, segmentation is completed, and a segmentation effect picture is output; if the objective function is larger than the iteration stopping threshold value, the novel membership matrix and the clustering center are used for calculating the quaternion distance and the objective function repeatedly until the objective function meets the iteration stopping condition, so that segmentation is completed. The image segmentation method based on quaternion space is provided on the basis of a traditional fuzzy C-means clustering algorithm, a channel R, a channel G and a channel B are effectively used as a unified whole to be considered, the integrity of color information can be effectively maintained, and segmentation results more accord with human eye vision.

Description

technical field [0001] The invention belongs to the technical field of image processing and computer vision, in particular to an image segmentation method based on quaternion theory and fuzzy C-means clustering FCM. Background technique [0002] Image segmentation is an extremely important analysis method in computer vision and pattern recognition. The purpose of image segmentation is to divide the image into several different, non-overlapping regions with unique properties, extract the objects of interest, and add a unique class label to each pixel. Image segmentation is an important part of image analysis, and has been widely used in medical imaging, face recognition, fingerprint recognition, traffic control systems and machine vision. [0003] The current mainstream methods of image segmentation include segmentation based on regions, segmentation based on random models, segmentation based on morphological theory, segmentation based on pixels, segmentation based on fuzzy ...

AI Technical Summary

Problems solved by technology

However, the color may not converge after pixel information mapping. If there is no obvious valley in the process of segmentation using the histogram technique, the segmentation effect will be seriously affected.

Images