Research On Color-based And Texture-based Image Retrieval Algorithm

With the development of modern network technology,image retrieval technology has become one of the hot topics.In the real life,due to the massive increasing of the image database and the perceived deviation of the human visual system to the image,and users need higher and higher requirements on the performance of the image retrieval technology as well,investigation on the image retrieval technology attracts more and more attention.In this thesis,we have analyzed and researched image retrieval techniques based on color and texture.In particular,the image retrieval algorithms based on color feature,texture feature and mixed of two feature fusion are studied,and some improved algorithms are provided.The main work in this thesis is as follows:(1)Color image retrieval based on color feature.In chapter three,we study the image retrieval algorithms based on RGB color space and HSV color space,based on which,a new image retrieval algorithm based on color feature is proposed.Firstly,the color feature of the image is extracted in the RGB color space and the HSV color space.Then,the normalization process is carried out.Finally,the Euclidean distance is used as the basis of image similarity measurement and image retrieval.The simulation results show that the system recall rate and the precision rate has been improved under the condition that two weights of the color feature are 50% and 50%.(2)Gray image retrieval based on texture features.In chapter four,we study the global texture information and local texture information of images,based on which,a new image retrieval algorithm is proposed.Firstly,the CS-LBP transform is used to extract the histogram features of the image.Then,the mean and variance of the image gray level co-occurrence matrix are calculated.Finally,the Euclidean distance is used as the basis of image similarity measurement,so as to do the image retrieval.The simulation results of the new algorithm show that the recall and precision are improved.(3)Image retrieval based on color feature and texture feature.In chapter five,we consider the color information and texture information of the image,based on which,a new image retrieval algorithm is proposed.Firstly,the color features of the image are extracted by combining two color spaces.Then,the texture feature of the image is extracted by CSLBP-GLCM algorithm.Finally,an image retrieval system based on color feature and texture feature is constructed.The simulation results show that the performance of the new algorithm is the best when the weights of the color and texture features are 60% and 40%,respectively.