Eigen-model based 3D model recognition from cylindrical mapped image

In this paper, we present a new method for 3D model recognition. Unlike other approaches, our model recognition is accomplished based on a plane which describes a 3D model and an eigen-model algorithm. By mapping a 3D model into a virtual cylinder around the model, we get the 3D model aspect information on an unfolded cylinder as a 2D image, which gives effective data format for feature extraction, selection and classifier design phases. We apply automatic optimal feature selection and a hierarchical database organization from the SHOSLIF algorithm to 3D object recognition.