基于NTSS和压缩感知的人脸跟踪算法

Face Tracking Algorithm Based on NTSS and Compressed Sensing

提出基于NTSS和压缩感知的人脸跟踪算法,采用不同的搜索半径和步长双重采样,运用NTSS搜索方法作为目标的检测方法。通过国际标准的人脸数据库AFW和MALF视频序列进行算法的测试,并使用中心距离均值算法对压缩感知算法和基于NTSS压缩感知的人脸跟踪算法进行比对,得出基于NTSS和压缩感知的人脸跟踪算法最大可降低59. 01%的误差;此外NTSS压缩感知人脸跟踪算法在人脸发生遮挡再跟踪、人脸几何变换和在暗光条件下具有较好的鲁棒性。

This paper proposes a face tracking algorithm based on NTSS and compressed sensing. It uses different search radius and step double sampling, and uses NTSS search method as the target detection method. The algorithm is tested by the international standard face database AFW and MALF video sequences, and the central distance mean algorithm is used to compare the compressed sensing algorithm with the face tracking algorithm based on NTSS compressed sensing. The results are that the face tracking algorithm can reduce the error by 59.01%. In addition, the NTSS compressed sensing face tracking algorithm has better robustness in face occlusion re-tracking, face geometric transformation and under dark conditions.