Mesh Animation Compression Via Frame-Clustering And Spectral Graph Wavelet

In recent years,3D mesh animation is applied on different fields such as3D online games and remote medical system. In order to show strong realistic sense, enriching the details of3D mesh animation constantly will bring exponential increasing data. Therefore, it has a great significance on the development and application of mesh animation to reduce the network bandwidth and storage space through designing various new animation compression algorithms. The compression of3D mesh animation is studied systematically in this paper, and the main work is completed as follows.Firstly, a kind of frame-clustering algorithm which can be used for progressive transmission of mesh animation is proposed in this paper. This paper not only uses the mesh residuals to replace the original Euclidean distance, but also add the key-time difference as a part of frame-distance. By that, we can ensure the index of frames in each cluster keep continuous.Secondly, this paper trying to use two ways to get signals which defined on meshes for spectrum wavelet compression:1) Select the representative frames from each cluster, and then compute the mesh residuals between the representative mesh and the other meshes;2) construct three trajectory matrixs in x,y,z direction respectively applied the PCA and obtain a set of coefficients which would be regarded as the mesh signal.Thirdly, a kind of spectrum wavelet transform defined on mesh directly used for mesh compression is introduced. Based on Spectrum graph theory and classic wavelet transform, this paper regards the above mesh residuals and PCA coefficients as the signal applied the spectrum wavelet transform. After that, quantize and encode the spectrum wavelet coefficients.Finally, above compression algorithm is realized by Matlab. It has obtained a good compression ratio after testing on several classical models. The comparison of several existing classical and newest related algorithm confirms the improvement of the algorithm proposed in this paper on reducing the store space and reconstruction error.