Multi-view ear recognition based on moving least square pose interpolation

• Main Authors: LIU, Heng, ZHANG, David, ZHANG, Zhiyuan
• Format: text
• Language: English
• Published: Institutional Knowledge at Singapore Management University 2009
• Subjects: Face recognition; move less square; small sample size problem; move less square approximation; Gauss weight function; Artificial Intelligence and Robotics
• Online Access: https://ink.library.smu.edu.sg/sis_research/7937; https://ink.library.smu.edu.sg/context/sis_research/article/8940/viewcontent/Multi_view_Ear_Recognition_Based_on_Moving_Least_Square_Pose_Interpolation.pdf
• Institution: Singapore Management University

Based on moving least square, a multi-view ear pose interpolation and corresponding recognition approach is proposed. This work firstly analyzes the shape characteristics of actual trace caused by ear pose varying in feature space. Then according to training samples pose projection, we manage to recover the complete multi-view ear pose manifold by using moving least square pose interpolation. The constructed multi-view ear pose manifolds can be easily utilized to recognize ear images captured under different views based on finding the minimal projection distance to the manifolds. The experimental results and some comparisons show the new method is superior to manifold learning method and B-Spline based recognition method.