Object-based Color Image Retrieval Using Concentric Circular Invariant Features

This paper proposes a novel object-based method using local concentric circular invariant features for color image retrieval. With this approach, the salient objects are firstly segmented and converted into several concentric circular subimages with the square-to-circular image transformation. The local invariant features for each subimage which are robust to image rotation, object translation and rotation are then extracted in a new polar coordinate space. Experimental results on three image datasets with different conditions validate that, compared with the traditional image retrieval methods, the proposed method can obtain high retrieval precision and speed simultaneously.