Study On The Extraction Of The Texture Feature In The Process Of Image Retrieval

With the development of the technology of multimedia and the digital library, content-based image retrieval has become a key problem of image processing and computer vision. Content-based image retrieval is a technique to retrieve related images according to their content information. Its goal is to provide a mechanism that can automatically recognize or understand important features of an image. This paper mainly studies the image processing technique based on texture feature, including the analysis of texture feature and the extraction of texture feature. At first, we introduce the current research situation of CBIR (Content-based image retrieval) both at home and abroad, basic theories, inquiry ways and application fields. Then we describe the definition of texture and texture feature in detail; introduce some usual methods to analyze the texture, including gray histogram, edge histogram, gray concurrence matrix, gray-gradient concurrence matrix and structure-based method; particularly describe Gabor filter, Gabor Wavelet and the method of Gaussian normalization, then deal Gabor Wavelet with Gabor Wavelet. In chapter 4, we describe the characteristics and classification of the similar measurement method, feedback technology in CBIR. At last, we design an image retrieval system and describe each module's function in detail. Comparison experiments of the method to other ones show that the efficiency can be improved 5-6 percent. This method can work better for the images which change visibly in the lightness, especially when the backgrounds are simple and the boundaries between the subjects and background are clear. We made a conclusion about the work presented in this paper and point out the emphasis of further research.