LOCAL GABOR TERNARY PATTERN BASED ILLUMINATION VARIABLE FACE RECOGNITION IN VIDEO

摘要

The illumination variation problem is one of the key problems for face recognition in uncontrolled environment. A method for face recognition to address the illumination variation problem is proposed, which combines Gabor filters with LTP operator. First, we convolve the image with Gabor filters to extract their corresponding Gabor feature maps, and then use the LTP operates on each Gabor feature maps to extract the local neighbor pattern. Finally, by using the histogram sequence extracted from all these region patterns, the input face image is fully described. The results compared with the published results on Yale-B and CMU PIE face database of changing illumination verify the validity of the proposed method.