Generic object recognition based on the fusion of 2D and 3D SIFT descriptors

This paper proposes a new generic object recognition (GOR) method based on the multiple feature fusion of 2D and 3D SIFT (scale invariant feature transform) descriptors drawn from 2D images and 3D point clouds. We also use trained Support Vector Machine (SVM) classifiers to recognize the objects from the result of the multiple feature fusion. We analyze and evaluate different strategies for making this multiple feature fusion applied to real open-datasets. Our results show that this new GOR method has higher recognition rates than classical methods, even if one has large intra-class variations, or high inter-class similarities of the objects to recognize, which demonstrates the potential interest of this new approach.