Sculpture Similarity Representation and Retrieval Application Based on Similarity Measurement Algorithm

Buddhist sculpture is an important part in culture heritage. How to process quantitative evaluation on artistic feature is a key point for further and precise analysis on art works. In this research, we proposed a digitalized approach for Buddhist head similarity evaluation and stylistic analysis. Our aim is to normalize the random 3D scanning data into a uniform structure and to extract global shape features. The data in the form of 3D descriptor are combined with the similarity measurement algorithm in stylistic retrieval process. By applying spherical harmonic degree balancing and principal components analysis on original coefficients, we get a simple and informative 3D shape descriptor. The experimental results demonstrate that our method is efficient, successful in Buddhist shape similarity representation, and insensitive to topological noise. Furthermore, based on our shape descriptor, the intrinsic relationship between shape features and historical environment were analyzed by clustering method.