Research And Implementation Of Texture-based Image Retrieval Technology

How to analyze, store and retrieve the huge amount of data efficiently and effectively, especially for those multimedia data is an imperative problem. This paper focuses on the texture-based image retrieval techniques. I conclude that image texture representation and image feature match should be two key research topics in the paper.Firstly, an overview of content based image retrieval and the characteristic and inquiry mode of it are presented. Then the research of algorithms of texture-based image retrieval is devised from the following three aspects:1. Texture-based image retrieval algorithms in spatial domain: perceptual-based Tamura texture feature is analyzed; Run-length texture analyzing algorithm is improved; Four laws' texture operators which perform better in texture analyzing are achieved.2. Methods in frequency domain: Through a comparative study, wedge and ring feature of Fourier transform performs better than other features based on texture spectrum. It represents the directivity, coarseness and regularity of an image.3. Methods in space/frequency domain: A group of Gabor filters are used in texture-based image retrieval; By analyzing how the energy changes with the decomposing of wavelet, tree-structured wavelet transform in texture-based image retrieval is selected to use.As for similarity measurement of texture-based image retrieval, This paper demonstrates two data models related to the similarity measurement and proposes two kinds of multi-feature combination structures. Furthermore, how to use relevance feedback in content-based image retrieval system is discussed. Some of the distance functions used in this paper are put forward in the paper too.This paper designs a CBIR system as the test bed for retrieval algorithms, which is an experimental frame system. By using a sorting assessment method, this paper assesses and compares the algorithms, The CBIR system proves to run well with a large image library. The developing tendency and the hotspot of research of the content-based image retrieval are expected in the paper.