3D Object Classification Based on Improved Interpretation Tree

In order to classify unknown 3D objects into a set of pre-determined object classes,the paper proposes a 3D object classification approach based on improved interpretation tree.In our approach,a set of novel integrated features and corresponding constraints are presented,which are used to define the efficient tree search rules,and obtain the feasible correspondences of object data and the stored models.A similarity measure computation algorithm is developed to evaluate the shape similarity of the correspondence.The classification approach can achieve different kinds of matches with the model similarity measures and the object class which the unknown object belongs to.The good performance of the proposed 3D object classification approach is demonstrated with a series of experiments.