Histogram of template for human detection
Shaopeng Tang, Satoshi Goto · 2010
In this paper, we propose a novel feature named histogram of template (HOT) for human detection in still images. For every pixel of an image, various templates are defined, each of which contains the pixel itself and two of its neighboring pixels. If the intensity and gradient values of the three pixels satisfy a pre-defined function, the central pixel is regarded to meet the corresponding template for this function. Histograms of pixels meeting various templates are calculated for a set of functions, and combined to be the feature for detection. Compared to the other features, the proposed feature takes intensity as well as the gradient information into consideration. Besides, it reflects the relationship between 3 pixels, instead of focusing on only one. Experiments for human detection are performed on INRIA dataset, which shows the proposed HOT feature is more discriminative than histogram of orientated gradient (HOG) feature, under the same training method.