Method for dividing level set image based on characteristics of neighborhood probability density function

Abstract

The invention discloses a method for dividing a level set image based on characteristics of a neighborhood probability density function, which mainly solves the problem existing in the prior method that the variability of texture elements of a natural image in texture scale and texture direction cannot be effectively processed. The method comprises the following steps of: initializing a level set function; setting an initial value and a maximum value for the iteration number of the level set function; working out a characteristic set of the neighborhood probability density function; working out characteristics of a subset probability density function; working out a similarity vector; according to an evolution equation of the level set, updating the level set function; judging whether to enter a cycle, if so, working out the characteristics of the subset probability density function, the similarity vector and the level set function again; otherwise stopping the cycle, and using a zero level set of the level set function as a segmentation border of an input image. Because the texture elements of the natural image are fit by a multi-scale sliding window, the variability of the texture elements in texture scale and texture direction can be better processed, so that the method can be applied to the segmentation of the natural texture image.

Description

technical field [0001] The invention belongs to the technical field of image processing, and in particular relates to image segmentation, in particular to a level set image segmentation method based on neighborhood probability density function features. The method can be used in the technical field of image segmentation. Background technique [0002] Image segmentation is the basis of image understanding and pattern recognition. It is widely used in medicine, military, meteorology, climate, environment and other fields. It is a hot and difficult point in current research. So far, there are many kinds of image segmentation methods, which can be generally divided into two categories: traditional image segmentation methods and image segmentation methods combined with specific theoretical tools. Traditional segmentation methods usually include three categories: boundary detection-based, region-based and a combination of the two. Theoretical tools often used in segmentation met...

AI Technical Summary

Problems solved by technology

However, for real natural texture images, the biggest deficiency of the existing level set image segmentation methods is that they cannot effectively deal with the variability of texture primitives in natural image textures in terms of texture scale and texture direction, resulting in the inability to accurately segment natural texture images.

Images