Research Of Image Retrieval Technology Based On Shape Feature

In order to take full advantage of the digital images, Content-based image retrieval (CBIR) technology emerged as the times require, which combined many fields of knowledge, such as image recognition, computer vision, database technology and artificial intelligence. At present, the main application is in professional library. Such as in trademarks' library, scanned newspaper library, tobacco leaves' library. Because shape is the stable characteristics of objects which does not change with the image colors and there is no exact definition of it, shape-based image retrieval becomes hot and difficult.The extraction and description of shape feature and similarity measures are two key issues of shape-based image retrieval, the paper studies them. Gallery of firearms is the researched object. In the extraction and description of shape, makes use of the technology of segment and morphology to get the object target, and calculates the Hu invariant moment's value of the target's region.The result shows that the morphology of the images which deal with internal and external reached a filtering effect. That connected to the brink of non-consecutive points. Weighted Euclidean distance lighted the impact of retrieval results, which the fluctuation of the higher-invariant moment's value.