ROI-HOG and LBP Based Human Detection via Shape Part-Templates Matching
Currently, Histogram of Oriented Gradient (HOG) descriptor serves as the predominant method when it comes to human detection. To further improving its detection accuracy and decrease its large dimensions of feature vectors, we introduce an improved method in which HOG is extracted in the Region of Interest (ROI) of human body with a combined Local Binary Pattern (LBP) feature. Via establishing human shape part-templates tree, a template matching approach is employed to improve detection results and segment human edges. The experimental results on INRIA database and images from practical campus video surveillance demonstrate the effectiveness of our method.
KeywordsPedestrian detection ROI-HOG descriptor LBP feature Shape part-template matching
Unable to display preview. Download preview PDF.
- 1.Dalal, N., Triggs, B.: Histograms of Oriented Gradients for Human Detection. In: IEEE Computer Society Conference on Computer Vision and Pattern Recognition, vol. 1, pp. 886–893. IEEE Press, New York (2005)Google Scholar