Multi-dimension texture image partition method based on self-adapting window fixing and propagation

Abstract

The present invention discloses a multi-scale grain image segmentation based on self-adaptive window fixation and spread. The process comprises the steps as follows: an image block n corresponding to the grain of an image to be segmented is picked up for wavelet transform, and a corresponding HMT model parameter theta n thereof is determined; the corresponding likelihood value of corresponding data block at each wavelet decomposition scale of the image to be segmented and the corresponding likelihood value of the pixel of the image to be segmented are determined respectively and are combined together to find out the likelihood value n<k> required by finial fusion; the likelihood value on fusion widest scale (k is 4) is found out, and the corresponding segmentation result plotting on the scale is determined also; a marking field on the fusion scale k and the physical clustering center of each grain are confirmed; the multi-scale segmentation of the self-adaptive window fixation and spread is used for find out the segmentation result plotting on next fusion scale k minus 4; the finial segmentation result is confirmed by judging whether the fusion scale of the segmentation result plotting is zero or not. The multi-scale grain image segmentation has the advantages of area consistency and good edge positioning performance and can be used for image segmentation comprising grain information.

Description

technical field [0001] The invention belongs to the technical field of image processing, in particular to a multi-scale texture image segmentation method. This method can be used in texture images and images containing texture information, such as synthetic aperture radar SAR, remote sensing images or medical image segmentation. Background technique [0002] There are a large number of things with certain texture characteristics in nature, such as large cornfields, forests, building areas, and so on. In practice, it is very important to divide and mark these areas for the layout and planning of the city, and texture image segmentation is the process of dividing the same texture area in an image. All segmentation methods aim to deal well with the identity and discontinuity in this process, that is, the contradiction between the uniformity of the region and the determination of the boundary between different textured regions. The method based on multi-scale thinking is simil...

AI Technical Summary

Problems solved by technology

[0005] These traditional segmentation methods based on multi-scale Bayesian fusion all use the connection between the class labels in the scale and the links between the class labels in the upper and lower scales to construct the context fusion background, and treat the connections between all the class labels in the same way, that is Using the SWAP method to obtain the weight of the influence of the upper scale on the lower scale cannot make full use of the multi-scale segmentation. The coarse-scale segmentation results have good regional consistency and the fine-scale segmentation results have good edge accuracy. It cannot solve the regional consistency well. The Paradox of Sex and Marginal Accuracy

At the same time, when this type of method constructs the context background, it only pays attention to the influence of the neighborhood information of the class label on the segmentation result, and does not pay attention to the aggregation characteristic of the texture image, that is, the texture belonging to the same class is a connected area, and there is a physical cluster center

Images