Image Retrieval Based On Shape Features

In recent years, with the emergence of multimedia technology and digital equipments, a great quantity of image information has swarmed into people's lives. How to effectively manage and access image information has become an urgent problem, which can be solved by a new kind of technologies, namely content-based image retrieval technology. The users of content-based image retrieval system prefer to search according to shape features in view of their more in line with the cognitive characteristics of images, and therefore, the shape-based image retrieval has attracted the attention of more and more researchers. However, the particularity of shape features accounts for the difficulty of realizing shape-based retrieval, which is the most challenging among the content-based image retrievals and of great significance for research.This paper first describes the background, current situation and future direction of content-based image retrieval, and classic CBIR system, analyzing and summarizing related knowledge on system structure, visual features, and feature matching and performance evaluation, especially discussing some key technologies on shape-based image retrieval in detail. Then on the basis of existing research results, is a new shape-based image retrieval method proposed, mainly including two core algorithms, namely the contour shape retrieval algorithm based on edge direction histogram and the region shape retrieval algorithm based on frequency domain. Finally, on the basis of algorithms mentioned above, the prototype system of image retrieval based on shape features is developed, with the realization of programming each module, and the comparison between retrieval performances of two algorithms.