Face recognition using locality sensitive histograms of oriented gradients

In this paper, we propose a novel locality sensitive histograms of oriented gradients (LSHOG) for face recognition. The traditional histograms of oriented gradients (HOG) divide an image into many cells, and computed a histogram of gradient orientations over each cell. Unlike traditional HOG our proposed LSHOG compute a histogram of gradient orientations over the whole image at each pixel location. For each occurrence of a gradient direction value we add a locality sensitive parameter which can make the gradient direction value declines exponentially in regard to the distance between the pixel location of the gradient direction value and the pixel location where we are computing the histogram. Our proposed LSHOG take spatial information of face images into account which make it insensitive to noise such as occlusion and non-uniform illumination. LSHOG was applied to face recognition task. First, we use LSHOG to extract feature vectors of face images. Then, to show how different dimension reduction algorithms affect recognition accuracy, we choose several typical dimension reduction algorithms to reduce the dimensions of feature vectors of face images. We evaluated LSHOG and our proposed face recognition method on four benchmark face databases. Experimental results verify the feasibility and effectiveness of LSHOG and our face recognition method.