Content-Based Feature Matching for Fragment Reassembly of Ceramic Reconstruction

A large amount of fragments of different ceramic objects are usually found during archaeological excavation. Finding the relation between these fragments is time-consuming and demands expert knowledge. In this paper, we study a related problem known as ceramic reconstruction. We present a novel content-based feature matching method and develop the ceramic fragment reassembly system (CFRS) for ceramic reconstruction. CFRS can automatically reassemble fragments to make it easier for human experts to further improve the quality of reconstruction if necessary. CFRS consists of three modules: fragment preprocessing, candidate matching, and matching-candidate selection. The fragment preprocessing module extracts features from images of fragments. Then, the candidate-matching module picks up matching candidates based on curve matching. Finally, in the matching-candidate selection module, Binary Particle Swarm Optimization (BPSO) is applied to select the final matching results. Experimental results indicate that the proposed method attains not only higher precision but also less computation time than the previous approaches.