The application of local features in 3-dimentional object recognition

This paper introduces the 3D object recognition method, which includes normal estimation, feature point selection, feature descriptor extraction, feature matching between the scene and the model in the library, hypothesis generation and verification. In this paper, we just focus on the first three parts, and use the analysis of the eigenvectors and eigenvalues of a covariance matrix and K-NN algorithm, NARF (normal aligned radial feature) and FPFH (fast point feature histogram) respectively.