Self-adapting multi-dimension veins image segmenting method based on wavelet and average value wander

Abstract

The invention discloses a self-adaptive multi-scale texture image segmentation method based on wavelet and mean shift, and relates to the image processing technical field. The invention aims to solve the contradiction between the region consistency of a texture image and the edge accuracy of adjacent regions in the regular method to achieve the effect that the image is effectively segmented under the condition of failing to obtain the priori knowledge. The realization process of the method is that: based on the orthogonal wavelet transformation, mean shift clustering without supervision is used to realize the segmentation of the feature of a wavelet transform coefficient on different scales; through the information transmission of the features among different scales, self-adaptive different regions as the image select appropriate segmentation scales, namely the interior of a texture region uses a coarse scale feature, the boundaries of different texture regions use a finer scale feature, thus the edge of the image is located more accurately while the region consistency is assured until the final segmentation result is obtained. The invention can be used to solve the texture image segmentation issue without any priori knowledge.

Description

technical field [0001] The invention belongs to the technical field of image processing and relates to the application of the technology in image segmentation, in particular to an adaptive multi-scale texture image segmentation method based on wavelet and mean shift. This method can be used to solve the problem of texture image segmentation without any prior knowledge. Background technique [0002] Texture image segmentation is a basic problem in image processing, and its main task is to divide an image into a limited set of regions with relatively consistent texture features. The two core issues it needs to solve are: the consistency of the texture area and the accuracy of the edge of the adjacent area. According to the basic characteristics of texture, the current texture image segmentation methods mainly include four categories: statistical method, structural method, model method and space / frequency domain joint analysis method. [0003] The statistical method considers...

AI Technical Summary

Problems solved by technology

[0006] The purpose of the present invention is: in order to overcome the deficiencies in the prior art, solve the contradiction between the consistency of the texture image area and the edge accuracy of the adjacent area in the conventional method, propose the method of the present invention, to achieve in the situation where prior knowledge cannot be obtained The purpose of effectively segmenting the image in the case of

Images