Support Vector Machine based 3D Object Recognition in a Virtual Environment

摘要

Support vector machine (SVM) represent a new approach to pattern recognition and has been shown to be particularly successful in many fields. This paper presents a nonlinear SVMs based approach for 3D object recognition in a vehicular virtual experiment environment. The system outline and 3D images recognition algorithm are depicted. By simulation experiment and the comparison of the results between the recognition accuracy of SVM and BP network, we illustrate the potential of nonlinear SVMs on images classification of different objects. The excellent recognition rates achieved in the performed experiments indicate that nonlinear SVMs are well suited for 3D images recognition, especially in coping with small sample sizes. This is vital for achieving 3D object recognition and human-computer interaction rapidly and accurately in virtual experiment environment.