摘要
研究了一种基于改进Camshift的人脸跟踪算法.该算法采用分块加权的直方图匹配方法增强人脸的辨识度,通过人脸形态约束筛选不合理的人脸形态变化,并结合Kalman滤波器预测修正跟踪人脸位置.实验表明,改进后的算法比经典Camshift算法有更强的抗肤色干扰能力和跟踪准确性.
The improved Camshift algorithm for face tracking is proposed, which enhances the identification of facial structure by using the block weighted color histogram in template matching, filtrates the unreasonahle facial morphological change according to the face shape detection, predicts and amends the accuracy of face tracking through Kalman filter. Experimental results prove the improved algorithm become stronger interfer ence ability, higher tracking accuracy than the classical Camshift algorithm.
作者
骆绍烨
刘丽桑
LUO Shaoye LIU Lisang(College of Information Engineering, Putian University, Putian 351100, China College o.f Information Science and Engineering, Fujian University of Technology, Fuzhou 350118,China)
出处
《延边大学学报（自然科学版）》
CAS
2017年第2期144-149,共6页
Journal of Yanbian University（Natural Science Edition）
基金
国家自然科学基金资助项目(81373552)
莆田市科技计划项目(2015G2014)
