ABSTRACT
In this chapter, the authors propose a methodology for watermarking as a means of content-based indexing and retrieval of three-dimensional (3-D) models stored in databases. They describe a novel technique for 3-D model indexing and watermarking, based on a generalized Radon transform. The authors present the proposed descriptor extraction method and the detailed watermarking procedure. They also describe the matching algorithm. Determining the similarity between 3-D objects is a challenging task in content-based recognition, retrieval, clustering, and classification. Further, the proposed watermarking method is robust to geometric distortions such as translation, rotation, and uniform scaling. Every time a model is used as a query model, the watermark detection algorithm is applied in order to retrieve the corresponding descriptor vector, which can be further used in a matching algorithm. Experiments were performed using two different databases to test both content-based search and retrieval, and watermarking efficiency.
