Analysis of 3D Textures Based on Features Extraction

This paper details an efficient method for the analysis of textures in the 3-dimensional space. The Local Binary Patterns (LBP) and the Grey Level Co-occurrence Matrix (GLCM) which are successfully used in various applications and the Decimal Descriptor Patterns (DDP) which is a new promising method are compared. The performance of these approaches is tested with a large number of 3D textures and evaluated in front of illumination variations, surface and texture rotations. Test of classification are performed over two 3D databases with the multiclass Support Vector Machines (SVM) classifier. Using the DDP method, excellent experimental results are obtained.