Image Retrieval Based On Adaptive Semi-Hilbert Curve

Content-Based Image Retrieval (CBIR) is an important and advanced technology in information and multimedia fields. In this thesis, a Adaptive Semi-Hilbert Curve has been proposed to overcome the shortcomings of the Standard Hilbert Curve. Furthermore, this new curve has been used in CBIR. The main contents and contributions of this thesis are the following:1. The Adaptive Semi-Hilbert Curve (ASHC) has been proposed to overcome the shortcomings of the Standard Hilbert Curve (HC). ASHC owns all virtues HC has. Furthermore, not like HC, which can only fill squares of certain sizes, ASHC can fill matrix of any size. That breaks the limits of applications of HC.2. The new curve (ASHC) has been used in CBIR. The concrete steps are as following. First, transform the image from 2-D discrete signal to 1-D discrete signal by scanning the image using ASHC. Second, carry out Discrete Cosine Transform (DCT) on the 1-D discrete signal. Finally, use the first several DCT coefficients as feature vector of the image. The retrieval method based on ASHC is a fast one.3. Analyze how quantification and number of DCT coefficients affect the results in the retrieval method based on ASHC.4. Experiments have been done to compare the retrieval method based on ASHC (ASHC-RM) and the retrieval method based on Global Color Histogram (GCH-RM). ASHC-RM has a better retrieval results than GCH-RM.5. Design and realize a simple retrieval system based on ASHC which has a friendly interface between computer and users.