Local histogram specification for face recognition under varying lighting conditions

Hui-Dong Liu,Ming Yang,Yang Gao,Chunyan Cui

DOI: https://doi.org/10.1016/j.imavis.2014.02.010

IF: 3.86

2014-01-01

Image and Vision Computing

Abstract:High frequency illumination and low frequency face features bring difficulties for most of the state-of-the-art face image preprocessors. In this paper, we propose two methods based on Local Histogram Specification (LHS) to preprocess face images under varying lighting conditions. The proposed methods are able to significantly remove both the low and high frequency parts of illumination on face images, as well as enhance face features lying in the low frequency part. Specifically, we first apply a high-pass filter on a face image to filter the low frequency illumination. Then, local histograms and local histogram statistics are learned from normal lighting images. In our first method, LHS is applied on the entire image. By contrast, in the second method, the regions contain high frequency illumination and weak face features on a face image are identified by local histogram statistics, before LHS is applied on these regions to eliminate high frequency illumination and enhance weak face features. Experimental results on the CMU PIE, Extended Yale B and CAS-PEAL-R1 databases demonstrate the effectiveness and efficiency of our methods.