Image Retreval And Application Based On Color And Shape Features

The rapid development of science and technology makes us walking in the information age in the ocean. Video chat, online shopping, remote control technology makes our life more convenient shortcut,at the same time, the digital information explosion at an alarming rate every day,vast amounts of data such as image, video, voice for our storage, management, access to information put forward unprecedented challenges. Digital image as a kind of rich information, intuitive and easy to understand information form, has the incomparable advantage over other information carrier. But with the increase of image information, image library scale constantly expanding, how to effectively manage and access to image information has become an urgent problem in the people. Content-based image retrieval technology arises at the historic moment, with its efficient retrieval speed and accurate query precision and become a hotspot in research of image retrieval in recent years, has a broad application prospect.Content-based image retrieval technology is by extracting image color, texture, shape and spatial relationship of low-level visual features to conduct a similarity measure for retrieval purposes.According to the query image retrieval in large image, realized based on image visual features of retrieval content,more accord with human visual perception.In this paper, on the basis of summarizes predecessors’ research results, the image of the color and shape features were studied, and achieve a more efficient multiple feature fusion image retrieval system. The system is divided into two stages. The first phase is at the beginning of the image library retrieval based on image color feature, determine the image class, thus narrowing the scope of retrieval. The second stage combines with the feature of the shape for secondary retrieval in one type of image, improve the retrieval accuracy. Such already contains multiple features fusion method on the retrieval precision of superiority,and further improve the retrieval efficiency. In terms of color extraction, the global color histogram and block color histogram, weighted processing, then it contains both the overall color distribution and the color spatial distribution information. Analysis comparison of the rectangular block with circle block, this paper proposes a circle block method such as area, made with the least amount of block number is contained as much as possible of subject image center area, save the storage space, reduced the amount of calculation. On the shape feature extraction, the image shape is described from two aspects:edge and region, analysis and comparison the Hu moment, Fourier descriptor, boundary direction histogram and so on the many kinds of shape feature extraction method. Simple effective boundary direction histogram method is used to obtain the shape information.In the end, through two stages to achieve the feature fusion of image retrieval based on color and shape, the simulation results verify the effectiveness of the algorithm in this paper.