CBIR (Content-Based Image Retrieval) is an important and advanced technology in information and multimedia fields. In this dissertation, a lot of research work has been done supporting CBIR with robustness. The main contents and contributions of this dissertation are on the following topic:(1) Build the image database and the feature database. By them, can contrast our approach and others.(2) Bring forward a novel way named Z image to settle the problem in color image processing and analyzing.(3) By using region description, the workload of match pixel by pixel will be avoided and the retrieval will be fast.(4) Taking advantage of the methods of moment, put forward a robust retrieval approach that is insensitive to illumination conditions, size and location of object.(5) Propose a non-geometrical similarity to satisfy the different requires of user.
|