3D Model Retrieval Based on 3D Fractional Fourier Transform

Liu YuJie1, Bao Feng1, Li ZongMin1, and Li Hua2
1School of Computer Science Communication Engineering, China University of Petroleum, China
2National Research Center for Intelligent Computing Systems, Institute of Computing Technology, Chinese Academy of Sciences, China

 

Abstract: In this paper, a new tool that is fractional Fourier Transform is introduced to 3D model retrieval. And we propose a 3D model descriptor based on 3D factional Fourier transform. Fractional Fourier transform is a general format of Fourier transform, and add a variables that is order. Our approach is based on volume. The first step of the approach is that voxelize these 3D models. A coarse voxelization is regarded as the input for the 3D discrete factional Fourier transform (3DDFRFT). A set of (complex) coefficient is obtained by 3DDFRFT in each order. The absolute values of coefficients are considered as components of the feature vector in each order. We also can integrate these feature vectors into the mixed feature vector, which is named as Multi-Order fractional Fourier Feature Vector (MOFFFV). We finally present our results and compare our method to 3D descriptor based on 3D Fourier Transform on the Princeton Shape Benchmark database.

Keywords: 3D discrete factional fourier transform, 3D model, feature extraction, retrieval.

 

Received September 23, 2010; accepted May 24, 2011

  

Full Text