Real-Time Human Detection Based on the Improved Local Binary Pattern

Abstract

By analyzing the detection accuracy and the testing speed of the Local Binary Pattern. we propose an improved LBP algorithm and apply it in human detection. Through the signs of the comparisons among neighboring pixels, it will get the histogram of the detection window. Then we can encode the global contour by the distribution coefficient of the histogram. when the Linear classifier is used, we propose a fast computational method that does not need to explicitly generate feature vectors and not require feature vectors normalization. experiment shows that this method has higher efficiency and can’t reduce the accuracy, it achieves 19 fps speed and can be used in a real-time system.