Multi scale block histogram of template feature for pedestrian detection

In this paper, a feature for human detection from still image is proposed. A multi scale block histogram of template feature (MB-HOT) is developed for human detection by extending the template level in the feature extraction. It integrates gray value information and gradient value information, and reflects relationship of three blocks. The feature is extracted from more macrostructures level and could represent more characteristic of human body. Experiment on INRIA dataset shows that this feature is more discriminative than other features, such as histogram of orientation gradient (HOG). Graphic process unit (GPU) based implementation is proposed to accelerate the calculation of two features, and make it suitable for real time application.