The Application And Study Of Ontent-Based Shoes Image Retrieval

With the development of the technology of computer multimedia and Internet, various and visual information is used more widely, shoes as indispensable supplies in people’s lives, their image data also risen sharply. As a result, effective methods of accessing and using shoes image data are urgent needed. Content-based image retrieval (CBIR) is a key technique to solve this problem. In this thesis, we discuss the application of content-based shoes image retrieval, extract the color and texture features from shoes image, and store in the database, then design a shoes image retrieval system based example of shoes image. This paper has mainly the following content:First, the color feature extraction of the shoes image. This paper extracts the color histogram of shoes image in HSV color space, the color histogram reflects the distribution of the proportion of the various colors in the image, and the color histogram is obtained by statistical color frequency of image. First, RGB color space was needed to convert to HSV color space, and the HSV color space was needed to quantify, and then calculate the proportion of each color component in the shoes image,72handle one-dimensional color histogram was obtained last.Second, the texture feature extraction of the shoes image. The paper selected extraction of shoes image based on the GLCM texture feature. Some statistics were extracted from shoes image’s Gray Level Co-occurrence Matrix and used to describe the image texture features quantitatively, including energy, contrast, entropy and correlation. This process is as follows:First, color digital images were needed to turn into grayscale images, and the gray level were needed to quantify, calculate GLCM of0degrees of the direction, then the statistics of GLCM were calculated, the parameters were as each texture feature vector component and normalized at last, Gaussian normalization method is used in this paper.Third, the system was realized based on the two above aspects, requirements analysis and system design of the system were described. The way of images retrieval is by example method. The system can display the search results visually, and the performance of this retrieval system was evaluated by recall and average integral ratio. The system can provide single feature retrieval of color or texture retrieval in this paper. Sort of two hundred retrieve images according to the similarity of size.