A Fast Method for Eye Location

摘要：

Based on AdaBoost algorithm and KLT corner detection algorithm, a novel algorithm for human eye location was proposed. First, cascaded face and eye detectors were trained by AdaBoost algorithm. Second, face regions were detected using the face detector combined with a priori rule of face edge. Third, the rough eye regions were extracted from the face regions using eye detector. Finally, the eye positions were located in the rough eye regions by KLT corner detection algorithm. The experiments on DTU_IMM test set result a correct rate of 99％ and a speed of 1 frame per second on 640*480 color images. Results show that the proposed algorithm is efficient.

关键词： eye location face detection AdaBoost,corner detection

作者: WU Tunhua WU Cong ZHOU Changle LIN Kunhui

作者单位: Department of Computer Science Xiamen University Xiamen, Fujian Province 361005, China Software School Xiamen University Xiamen, Fujian Province 361005, China

会议类型: 国际会议

会议名称: 第二届国际计算机新科技与教育学术会议(Proceedings of the Second International Conference on Computer Science & Education ICCSE2007)

会议地点: 武汉

会议语种:英文

页码: 636-639

在线出版日期: 2007-07-25（万方平台首次上网日期，不代表论文的发表时间）