基于快速SIFT算法和模糊控制的人脸识别

Face recognition based on fast scale invariant feature transform algorithm and fuzzy control

针对传统的人脸识别系统在复杂背景情况下不能实时准确进行人脸识别的问题,提出一种基于快速尺度不变特征变换(SIFT)算法结合模糊控制的人脸识别方法。首先,由SIFT特征点子区域方向直方图计算得到4个新角度,代表特征点方向信息。然后,在特征匹配阶段,根据SIFT特征点角度信息以及大小限制特征点匹配范围,简化算法复杂程度,得到快速SIFT算法。最后,引入闭环模糊控制系统,减少SIFT特征误匹配,提高人脸识别率。实验结果表明:基于快速SIFT算法的人脸识别方法平均识别时间提升了40%,在发生光照、姿态、表情等均有变化的复杂环境下人脸识别精度提高10%。

Most traditional face recognition systems can not be implemented for fast and accurate face recognition under cluttered background.To solve this problem,an improved Scale Invariant Feature Transform（SIFT）method and a fuzzy control strategy are proposed.First,four new angles are computed from the sub-region orientation histogram,which represent the orientation information of each SIFT feature.Then,the progress of face recognition is limited in a range based on the new angles,meanwhile the SIFT features are split into two types according to the size;only the features of the same type are computed,leading to significant simplification of the algorithm,thus a fast SIFT algorithm is obtained.Finally,a fuzzy closed loop control system is applied to increase the accuracy of face recognition,which leads to a decrement of the incorrect matching.The results show that the computing speed of the improved SIFT method is raised more than 40% comparing with the original SIFT algorithm and the recognition rate is raised 10% even under the clutter conditions where theillumination,posture or expression are changing.