Level Set Image Segmentation Based on Non-independent and Identically Distributed

Abstract

Aiming at the problems of traditional image segmentation methods such as single image type and falling into false boundary, a level set image segmentation algorithm based on non-independent and identically distributed (non-IID) is proposed. The traditional image segmentation method basically assumes that all samples are independent and have the same distribution. However, this assumption does not hold. In this paper, the image is segmented into several pixel blocks by super-pixel segmentation method. The feature vectors are extracted from the pixel blocks by non-IID method. We combine the obtained feature vectors with the CV model to change the energy function of the level set method and apply the new energy function to image segmentation. The experimental results show that the proposed algorithm can segment all kinds of images and has a good segmentation effect for weak edge images.