An Automatic Chinaware Fragments Reassembly Method Framework Based on Linear Feature of Fracture Surface Contour

Abstract

For Chinaware fragments, it is difficult to assemble them directly without considering the wonderful patterns painted on them. Given the simplicity of the Chinaware designs, each object contains similar textures and patterns. Compared to the oddly diverse appliance modeling, the difference between fragments of different Chinaware is small. The fragments are small and the fracture is flat, and there are many potential matching objects. For the Chinaware fragments’ reassembly, most of the work is still done manually. There is little available fully automatic reassembly work, and these approaches are for the reassembly of a single Chinaware. But what reality demands is the reassembly of the multi-Chinaware. Therefore, this article proposes an automatic method, whose strategy is analogous to the manual assembly, to accomplish this complex task. First, segment the contours of fracture surfaces of the fragments by the geometric feature of fracture edge; then, the contours are matched using our proposed multi-scale linear feature descriptor; given the massive fragments of multiple objects, we use the inaccurate matching strategy to build a matching set for each fracture surface contour and perform contour matching in this set. In this article, the fracture surface is segmented using the 2D slope information on the fragment edge. The descriptor proposed in the article uses distance triangle towers and chained angles to describe the “undulations” of the fracture surface. Moreover, the article uses a strict absolute-advantage-principle to reject false matching. In addition, after the initial reconstruction, we will iteratively adjust fragments according to the gap between the fragments to achieve better multi-fragment matching results. In this article, 18 porcelains, a total of 103 pieces, have been tested. Experiments are also carried out for special cases, including fragment reassembly with missing fragments and fragment reassembly with redundant fragments. The experimental results showed the effectiveness of method. However, the types of experimental data we currently have are relatively simple, and there is no way to reassemble the fragments in minutes. We hope to enable faster reassembly of more fragments in the future.