Infrared Versus Visible Image Matching for Multispectral Face Recognition

Multispectral face recognition has been an interesting area of research where images obtained from different bands are matched. There are many face image datasets available which contain infrared and visible images. In most face recognition applications, the IR image taken in different circumstances is matched against the visible image available in the application database. High computational cost is required for processing these images. In the literature, there is no guideline about the optimal number of features for dealing with multispectral face datasets. Thus, in this paper, we will perform image matching using infrared and visible images for face recognition and establish the threshold of the optimal number of features required for multispectral face recognition. The experiments conducted are on SCFace-surveillance cameras face database. The experimental setup for multispectral face recognition using LBP and PCA feature sets and the experimental results are discussed in the paper.