The Research About Image Retrieval Based On Texture Feature

Content-Based Image Retrieval(RGB) is valued by more and more professors because it has the advantage of high-speed,conventent and accuracy.It is the one who analyses the picture and extract its features and establishs the database by using the features and retrieves the picture's informations in the database,last,acquiring the pictures which the user needs. Texture is an important spatial feature useful for identifying objects or regions of interest in an image,so texture feature extraction is an important part in content-based image retrieval system. Nowadays,more algorithms abourt content-based image retrieval based on texture have applied,but the accuracy of the texture and the its use in the system can't meet people's needs.First,gray level co-occurrence matrix features extraction,Tamura texture analysis and Gabor wavelet is researched when the ways of texture feature extraction is researched.Soon afterwards,retrieveing the database established based on the various ways.Last,precision and recall is as the standard to compare the result based on the various ways.Texture feature extraction using gabor wavelet transform has the better consequent which is proved.Second,a novel algorithm of texture feature extraction based on the Fourier transform is proposed for improving efficiency of texture feature extraction. First of all,the picture is transformed by Fourier, in the frequency region,12 sectors is used and extract texture features, average and standard of the region are as the features.Last, Caneberra distance is used for estimating the two images similarity and the result is exported form small to large. So the arithmetic has higher efficiency and better exact than other methods on retrieval texture features.Last,designing system of picture's retrieval based on the knowledge of the color histogram in the space of HSV and the LBP.The system achieves the function which can rerieve the pictures based on the color and texture.It uses the Visual C++7.0 as the development tool and using MFC fram to programming,the system has the basic friendly interface.Later,the algorithm based the chapter compared with the other methods which uses single features,the result proves combining the features will improve the efficiency of the retrieve the pictures.Having an initial understanding for designing the system and having a better usderstanding for the practical application of C++ language when deviseing the system in the chapter.