| Bronzes are one of the Chinese cultural treasures.A large number of broken bronze accumulated with the rapid development of the Chinese cultural heritage undertakings,and traditional technologies are mainly used in the restoration of bronze,however.It is not only difficult to ensure the quality of restoration,but also time consuming,with the traditional technology of bronze restoration.Therefore,the only way to solve the increasingly prominent problem of bronze restoration is to combine the modern digital technology with the traditional technology to develop a new restoration technology.The patterns and inscriptions on the bronze surface can be considered as bas-relief features,and it is required to ensure the continuity of the patterns and inscriptions during the restoration of bronzes.Thus,it is important to investigate relief extraction for the digital restoration of bronzes.The goal of this thesis is to investigate relief extraction methods for the triangular meshes to lay a foundation for digital restoration of bronze s.The main research contents and contributions of this thesis are listed as follows:(1)An algorithm for relief extraction based on 3D surface rolling guidance is implemented.In this algorithm,the relief is considered as undulation or embedded parts attaching on the base surface.The spatial position of the base surface is reconstructed by estimating the normal vector of the base surface accurately.Then the distance between the relief and the base surface is calculate d for each point.Finally the relief is filtered out with a clustering algorithm.Experimental results show that a high-quality base surface can be generated,and reasonable extraction result can be obtained with the algorithm for models dominated by base surface.(2)An algorithm for relief extraction based on height estimation is implemented in this thesis.In this algorithm,a relief surface is considered as a surface composed of two components,a base surface and a height function which is defined over this base surface.Since the base surface on which the relief is located is unknown,it is challengeable to analyze the height function.A robust height function over the base can be estimated with the algorithm without extracting the base surface explicitly.Then a Gaussian mixture model(GMM)is employed to analyze the distribution of the height values to find the threshold automatically.Experimental result s have justified that the algorithm is effective for high relief.(3)An algorithm for extracting sharp features of the relief based on feature enhancement is proposed and implemented in the paper.Firstly,a weighted projected-distance is proposed to eliminate the adverse effect of the non-uniformity of the mesh on the feature extraction.Then,reverse bilateral filtering is proposed and utilized to generate an unsharp mask to enhance the projected-distance.Finally,Otsu’s method is applied to the histogram of the projected distance in order to determine the optimal threshold automatically,which is used to detect potential sharp features.Various experiments show that this method can extract the sharp features of the relief automatically and efficiently,especially for the bronzes with complex patterns. |
