基于图像特征融合与决策融合的多模式人脸识别方法

利用可见光图像和红外热图像进行图像融合是多模式人脸识别领域的一个新的研究方向．分别从特征级和决策级两个层次上研究了可见光图像和红外热图像的融合问题．在特征级上，引入遗传算法进行特征的优选，实现了两种图像的特征融合；在决策级上，提出利用Dempster-Shafer证据理论来实现决策的融合，并给出了具体的融合方案．分别采集了50人的红外热图像和可见光图像，每种各10张，共1000张图片进行了实验研究．实验结果表明，无论是对两种图像进行特征级融合还是决策级的融合，融合以后最终得到的识别准确率都大大提高，对于LDA和D_LDA方法达到了100％的准确率因此，可以认为基于遗传算法的特征融合方法和基于Dempster-Shafer证据理论的决策融合方法是实现多模式人脸识别的可行方法．

Multimodal Face Recognition Based on Images Fusion on Feature and Decision Levels

Face recognition based on image fusion from visual and infrared images is a new study focus of the multimodal face recognition. In this paper the fusions of visual and infrared images on feature level and decision level were discussed. On the feature level the feature fusion was realized due to the effective dimension reduction using genetic algorithm(GA); on the decision level the fusion method based on Dempster-Shafer evidence theory was proposed. Totally 1 000 pictures of 50 subjects were taken for the f...