Entropy Coding Optimization And Software Implementation Of FAMC Encoder For Animated 3D Mesh Compression

The digital Tri-Dimensional (3D) graphics data is used more and more widely in people’s social life.3D models play an increasingly important role in many fields, such as product manufacturing, construction designing, public entertainment, education, etc. With the constant emerging of 3D modeling tools and gradual maturity of 3D scanning technology in computer graphics, people put forward to higher requirements in the precision and details of 3D models. With the aid of advanced modeling tools and 3D scanning devices, the generated 3D models are more and more complicated in the structure, which leads to the rapid increasing of the amount of 3D graphics data. Therefore, how to effectively compress the huge data of 3D graphics is urgently to be solved.Frame-based Animated Mesh Compression (FAMC) has achieved excellent performance within the area of 3D graphics compression. The main work of this paper lies in the implementation of the FAMC encoder. Based on the FAMC compression framework and different functionalities, FAMC is divided into four modules, which are motion estimation, transform & quantization, transform domain prediction and entropy coding. By applying 1D transform in the temporal direction, transform module reduces the residual temporal correlations within the prediction error signal. Transform dimension increases over time, which makes it possible to improve precision of 3D Mesh step by step, which is called quality scalable coding. Transform domain prediction module supports vertices layering, firstly by the vertices of the first floor build the basic framework of 3D graphics, followed by the second and the third layers of multilayer vertices accumulate, which gradually improve the accuracy of 3D graphics. This is called spatial resolution scalable coding. In addition, hierarchical B frame structure allows the FAMC encoder to discard the B frames if necessary. This is called temporal resolution scalable coding. In summary, FAMC supports progressive transmission. Compared to single rate of compression transmission mode, FAMC greatly improves the experience of customers.Entropy coding is an important part of FAMC encoder and FAMC adopts context-based adaptive binary arithmetic coding (CABAC) as entropy coding tools. Conventional entropy coding of transform coefficients in FAMC has some deficiencies, for example the coding efficiency of the significance map is low and also the original FAMC has to repeat the process of coding amplitude information of transform coefficients for selecting the best encoding parameters, which greatly increases the computation complexity of the FAMC encoder. The binarization results of transform coefficients still remain statistical redundancy, which can be used to further increase the coding efficiency. With the above analysis, this paper proposes an improved coding scheme for significance map, which can reduce the coding bits of location information. In terms of coding the amplitude information of transform coefficients, this paper puts forward an improved binarization scheme that can adjust encoding parameters adaptively according to the transform coefficients value. The proposed scheme reduces the computation complexity and improves the coding efficiency. In addition, the improved context modeling method allocates appropriate context templates for the codeword suffix and the regular coding mode is applied to the codeword suffix, which increases the coding efficiency.