Image texture classification method based on shear wave and Gaussian mixture model

Abstract

An image texture classification method based on a shear wave and a Gaussian mixture model comprises the steps of performing directional multi-scale decomposition on a given training sample set with the same class number by means of the shear wave, acquiring sub-band coefficients of the training samples, and then constructing energy characteristics of the sub-band coefficients; representing the directional sub-bands by means of the calculated energy characteristics, and performing dimension reduction processing on the energy characteristics according to a kernel principal component analysis (KPCA) method; establishing a Gaussian mixture model (GMM) of the energy characteristics, and estimating a parameter of the Gaussian mixture model by means of an expectation maximization (EM) algorithm; and finally performing texture image classification by means of a Bayes classifier. The image texture classification method has beneficial effects of effectively improving classification precision of the texture image, effectively adapting with the texture image with a relatively small dimension, sufficiently utilizing directional multi-scale information of the texture image, and realizing high application value.

Description

technical field [0001] The invention relates to texture image classification in the field of pattern recognition and computer vision, in particular to an image texture classification method based on shear wave and Gaussian mixture model. Background technique [0002] Texture exists widely in the objective world. It is a basic attribute to express the surface or structure of objects, and it is also a very important research direction in computer vision. The intuitive meaning of texture is very clear. But people still have a vague idea of ​​what texture is. In graphics processing, texture has a broad and general meaning. Texture-based analysis and applied research has been conducted for nearly six decades. The research on texture is still very active. Many research institutions at home and abroad are engaged in this work. In recent years, in internationally renowned magazines and important conferences, research results on texture recognition emerge in endlessly, as many as ...

AI Technical Summary

Problems solved by technology

[0003] The technical problem to be solved by the present invention is to provide an image texture classification method based on shear wave and Gaussian mixture model, which is used to solve the problems of low classification accuracy, large feature dimension and time-consuming processing of pictures in existing classification methods

Images