带反馈的决策融合在人脸识别中的应用

Application of Decision Fusion With Feedback in the Face Recognition

将多传感器决策融合理论应用于人脸识别,并兼顾决策融合对通信资源要求低的优点和信息损失量大的缺点,提出一种基于支持向量机和D—S证据理论的带反馈的决策融合算法,不仅克服了目前单摄像机人脸识别在用户不主动配合、姿态变化和光线变化等环境下受到的严重限制,而且提高了人脸识别的准确度和可靠性。最后,对投票表决法、不带反馈的决策融合算法及本文算法在自采集的同一时刻多角度人脸库上的实验结果进行比较和分析,表明了本文算法的有效性和合理性。

Muhi-sensor decision fusion theory is applied to face recognition.Decision fusion needs the lowest communication resource, but has the largest information loss. In this paper, a decision fusion algorithm with feedback based on SVM and D-S is presented to overcome the severe limitations of single-camera face recognition when the subject is not cooperative, or there are pose changes and different illumination conditions,and achieve higher veracity and reliability.Finally, an experiment is taken on my face database that is muhi-view and simultaneous,to compare it with voting algorithm and the decision fusion algorithm without feedback information to verify its rationality and validity.