Pedestrian detection using an RGB-depth camera
Wen-Chang Cheng · 2016
This paper proposes a real-time pedestrian detection system using an RGB-D camera. The system takes gray-scale image and depth image as the system input, and uses the sliding window of fixed size to extract the candidate area. Then it extracts HOG features for the candidate area. Finally it conducts feature vector classification using AdaBoost classifier to finish pedestrian detection. The experimental results show the detection rate using gray-scale image and depth image is quite close to that using gray-scale image only, which would get relatively low false alarm rate. Therefore, the overall accuracy is improved to achieve practical purpose.