A novel direction chain code-based image retrieval

The direction chain code has been widely used to encode the boundary lines for its simplicity and low storage requirement. One problem with the chain code is that it lacks of robustness to the rotation and scaling of the image content, which limited its use in image retrieval. In this paper, we propose a novel shape descriptor called minimize sum statistical direction code (MSSDC), which has the advantages of being invariant to the position and rotation of the image content. It is proportional to the image scaling. A new distance measure called direction entropy (DE) is presented to measure the similarity of shape information embedded in the MSSDC. Experiment demonstrates the efficiency of our proposed algorithm.