Abstract: To get the good performance of infrared face recognition from the biological feature and Combine the local and whole character, a novel method based on based for infrared face recognition was developed. A new method based on FLD for feature extraction was presented combing blood perfusion and block DCT in wavelet domain was proposed.Firstly,thermal images were converted into blood perfusion domain by blood perfusion model to obtain consistent facial images without effect of ambient variations. Secondly blood perfusion data were decomposed using two scales' discrete wavelet transform. hhen, the component of low frequency sub-bands was partitioned into sub-blocks, to which the DCT was further applied. The FLD was applied to the global features combined by the extracted coefficients from all subblocks in DCT domain. Finally, Euclidean distance and the 3-NN classifier were utilized in recognition. The experiments conducted illustrate that the method proposed in this paper has better erformance compared with traditional PCA,PCA + FLD in thermal images.

Key words: FLD,Blood perfusion, DCT, Infrared face recognition