An Earthenware Reconstruction Method Based on the Matching of Both Contour Curve and Color Pattern on the Surface of Potsherds

Abstract

An earthenware reconstruction system, which can automatically reconstruct earthenware from given potsherds has been developed, in order to support the work of archaeologists. The existent system extracts the contour from given image of potsherd, and detects a candidate of joint by comparison of their contour shape. In this research, we utilize their color and surface pattern as clues in reconstructing earthenware as archaeologists do. This paper describes a method that extracts the feature of color and shape from input images, and proposes an effective algorithm of joint detection, which utilizes both color and shape feature. We have implemented the algorithm into the earthenware reconstruction system, and also report good performance results for excavated potsherds.