A Retrieval Method of Medical 3D Models Based on Sparse Representation

To assist physicians to find the required medical 3D models from a large-scale 3D model library, we propose a new retrieval method based on the sparse representation and regular relaxation method. In this method, the feature vector library of the 3D model database is considered as a set of coordinate basis. In order to get the model database coordinates of the retrieved model, the feature vector of the retrieved model are projected to this coordinate basis. So, the values of the coordinates are the retrieval result, because most values of the coordinates are 0. In the process of processing, we designed a new objective function to fit 0-norm and to avoid solving NP problems and unsolvable situations. The experimental results show that the proposed retrieval method has high precision, strong stability, and is suitable for handling large-scale data.