A Method for Robust Multispectral Face Recognition Francesco Nicolo and Natalia A. Schmid West Virginia University, Department of CSEE Morgantown, WV, USA 26506-6109 fnicolo@mix.wvu.edu,Natalia.Schmid@mail.wvu.edu Abstract. Matching Short Wave InfraRed (SWIR) face images against a face gallery of color images is a very challenging task. The photometric properties of images in these two spectral bands are highly distinct. This work presents a new cross-spectral face recognition method that encodes both magnitude and phase of responses of a classic bank of Gabor filters applied to multi-spectral face images. Three local operators: Simplified Weber Local Descriptor, Local Binary Pattern, and Generalized Local Binary Pattern are involved. The comparison of encoded face images is performed using the symmetric Kullbuck-Leibler divergence. We show that the proposed method provides high recognition rates at different spectra (visible, Near InfraRed and SWIR). In terms of recognition rates it outperforms Faceit R ⃝G8, a commercial software distributed by L1. Keywords: Face recognition, SWIR, Gabor wavelets, Simplified We- ber Local Descriptor, Local Binary Pattern, Generalized Local Binary Pattern, Kullback Leibler divergence 1 INTRODUCTION Face recognition has been an active research topic since 1990s. Different spec- tral bands of electromagnetic spectrum such as visible, Near Infra Red (NIR) (750nm − 1100nm) and thermal Infra Red (7 − 14μm) have been used to col- lect images and videos of people for testing various face recognition approaches. In majority of cases, these approaches are designed to perform face recognition within one specific spectral band. However, this scenario does not often support modern face recognition applications. Surveillance cameras, for example, often operate in both visible and NIR bands and switch between the bands depending on the night or day environments. Since gallery and watch lists are traditionally composed of visible light face images, newly designed face recognition methods are expected to be successful in matching NIR data versus color images. Apart from this case of cross-spectral comparison, attention has been re- cently turned to previously unexplored part of the Short Wavelength IR spec- trum (1100nm − 1700nm). The SWIR band has several advantages over the NIR spectrum. SWIR imaging modality produces clear images in the presence of challenging atmospheric conditions such as rain, fog and mist. It also works