Pedestrian detection by PCA-based mixed HOG-LBP features

As the basic method of feature extraction, HOG and LBP are paid much attention in the field of pedestrian detection. But the HOG features are too narrow and complex, so we simplify the HOG from the two perspectives. After that, PCA is used to reduce the dimensions of simplified HOG features and LBP features and then combine the two features after the dimensionality reduction to extract the mixed HOG-LBP features, and then used for pedestrian detection. Compared with conventional HOG, The experimental results show that PCA-HOG-LBP features reduce the dimensions and complexity of operator; improve the speed and accuracy detection of algorithm.