Multi-spectral information regarding face recognition

摘要

Recently, hyper-spectral facial image capturing techniques have opened a new door for creating innovative techniques aiming to improve these systems features. In our work, we have developed a verification approach based on Principal Component Analysis, and doing channel fusion (data fusion), in particular 5, 13, 14, 16, 24, 33, and 34 channels, reaching an Equal Error Rate of 97.37%; showing a good behavior of the kind of images.