The 3D Model Compression Method Based on Sparse Representation

The surface of the object is continuous and smooth. As one of the most popular methods, polygon meshes are extensively used in computer graphics. But it is difficult to achieve the sparse representation of polygon based 3D model due to its raw and discrete characters. Thus, the compressive sensing based compression method cannot be applied on 3D model directly. However, the geometry images method provides a way for the sparse representation of 3D geometric data in advance of its regular structure. For the sparse representation, we first proposed a new normalization method for the geometric patches of the geometry image. Then, a over-complete dictionary based sparse representation method for normalized geometry image has been proposed. The measurement matrix was introduced to sampling the geometric data which can be sparse by over-complete dictionary. At the decoding, we can get the reconstructed geometric data by solving the optimization model. Our effectiveness results of this method are shown in the experimental section.