Content-Based Image Retrieval using Shape Information of Central Object

Images in a database have different sizes, which may not be processed equally. In addition, if normalizing the sizes uniformly, it would inconveniently need another pre-process prior to extracting the characteristics. The study uses the object definition of image and color correlogram in order to re-establish the image retrieval area as an objective area. In addition, to solve deteriorated performance due to color retrieval, the image shape info is used. The defined area is not significantly influenced by background color, so the proposed method may have better results in the retrieval priority.