Accurate eye localization in the Short Waved Infrared Spectrum through summation range filters

•Geometric normalization is critical to most face recognition (FR) algorithms and is usually based off eye locations.•Our proposed approach quickly and accurately determines the pupil locations across the SWIR band.•Pupils are found using normalized correlation coefficients and summation range filters.•It is robust to image degradations such as blur, image downsizing and image compression.•The eye locations obtained from the proposed algorithm out perform other tested algorithms in both eye detection and FR results.