Hardware design of histograms of oriented gradients based on local binary pattern and binarization
Shen‐Fu Hsiao, Chan Jun-Mao, Ching-Hui Wang · 2016
Histograms of oriented gradients (HOG) are widely used to extract image features for pedestrian detection, but the involved complex computations are not amenable for hardware implementation. This paper presents a simplification of HOG using local binary patterns for gradient and magnitude computation. Furthermore, the complex normalization is replaced by simple binarization, leading to significant saving of area cost without sacrificing too much detection rate.