Semi-supervised multi-target clustering image segmentation method based on Chebyshev distance

Abstract

A semi-supervised multi-target clustering image segmentation method based on a Chebyshev distance is disclosed. The method comprises the following steps of marking a data sample point; then, calculating a corresponding class mean value of the marked data sample point, carrying out semi-supervised multi-target evolution clustering based on the Chebyshev distance on color image data, and using a Minkowski score index to select one optimal individual from an acquired approximate pareto optimal non-domination solution set; and according to the optimal individual, carrying out category classification on pixels in color images and acquiring a final segmentation result of the images. By using the method, different clustering criterion functions can be optimized simultaneously and a local optimum condition is not easy to generate. The Minkowski score index is used to select the appropriate optimal individual from the finally-acquired non-domination solution set according to supervision information and then an ideal image segmentation effect is acquired so that targets can be segmented completely. In the invention, the supervision information can be effectively used to acquire the ideal image segmentation result and algorithm realization is simple.

Description

technical field [0001] The invention belongs to the field of image processing, and in particular relates to a Chebyshev distance-based semi-supervised multi-target clustering image segmentation method. Background technique [0002] Image segmentation is an important research content in the fields of image processing, pattern recognition and computer vision. It divides the image into several specific regions with unique properties according to the color, texture, grayscale, etc. of the image and extracts meaningful targets. out of the technology and process. The quality of the image segmentation results directly affects the subsequent feature extraction, image recognition, understanding, and the performance of the visual system. Although a large number of various types of segmentation algorithms have been proposed so far, it is still very difficult to completely segment the target from the background information of the image due to the complexity and correlation of image inf...

AI Technical Summary

Problems solved by technology

Although a large number of various types of segmentation algorithms have been proposed so far, due to the complexity and correlation of image information, it is still very difficult to completely segment the target from the background information of the image.

Images