XOR 연산 기반 예측부호화를 이용한 노말 데이터 압축

Abstract

The well-known 3D mesh compression standards such as MPEG-4 3DMC provided a good compression gain of 3D mesh model. The standard compression technologies are highly efficient in compressing connectivity and geometry data. For other attributes like surface normals, the compression efficiency of the standard compression technologies is not as high as connectivity and geometry. This is because they do not exploit the characteristics of the individual attributes for further compression. Especially, surface normals are very hard to compress because it is hard to design an efficient prediction of the current normal from the previous (or neighboring) surface normals. In this paper, we proposed an efficient surface normal data compression method based on XOR operation. We found that the output values of XOR operation between the current and previous surface normals are highly correlated. And entropy coding of the output values of XOR operation yields the better compression gain than the existing methods. Experimental results showed that a compression gain between 19 to 38 percent is achieved using the proposed method. We suggested proposed method into 3D Mesh Coding (3DMC) standardized in MPEG-4 due to upgrading the algorithm with very small complexity because 3DMC spent most of time to compress geometry and connectivity.