基于血流图DCT域PCA和FLD的红外人脸识别

Infrared Face Recognition Based on Blood Perfusion Model PCA and FLD in DCT Domain

基于图像压缩思想及实际应用的考虑,提出一种基于血流图DCT域PCA和FLD相结合的红外人脸识别方法.根据生理学知识及生物力学的原理,把人脸的温谱图转换成血流图,通过DCT变换对人脸图像进行压缩,使变换域的能量集中在低频分量附近,从而减小了数据量,用主成分分析(PCA)和Fisher线性辨别分析(FLD)来提取人脸特征,通过三近邻分类器得到最终的识别结果.实验结果表明,本文的方法可以节省大量的存储空间和减小算法运算时间,并且在小样本集的情况下,也能取得较好的识别性能.

Infrared face recognition based on blood peffusion model and combination PCA with FLD is proposed in this paper,which consideration of image compression idea and the practical application. Firstly, Each infrared face image is first converted into blood perfusion model that is vulnerable to the external environmental was converted into more stable blood data, and then using DCT transformation for image compression, which make the energy concentrate in the vicinity of low-frequency components, thereby reducing the amount of data, then extracted facial features by PCA and FLD. Finally, Euclidean distance and the 3-NN classifier are utilized to obtain the last recognition results. The large number of experiments has demonstrated that the method proposed in this paper can reduce the storage requirement and the time complexity of computation and has better recognition performance in small sample set. It is shown that our proposed method is efficient and stable for application.