Influence of binning in Histograms of Oriented Gradients method representation

This paper focuses on comparing orientation binning to perform Histograms of Oriented Gradients algorithm for the feature extraction process. The HOG algorithm comprises of gradient computation, orientation binning and block normalization aimed towards object detection. Three angles of binning are considered for (3bins, 6bins, and 9bins) with angles.64×128 image sample is taken from INRIA Person Data set to extract the feature points, using which human detection can be enhanced. The HOG technique is observed through various orientation binning obtained from experimental simulation results on MATLAB.