Research on Human Body Recognition and Position Measurement Based on AdaBoost and RGB-D

Abstract

Recognition, detection and localization of the human body have been playing an important role in the development of robotics and interaction between machines and humans. But the speed and accuracy of target recognition and positioning are still not satisfying. To accurately identify the target body and measure the distance, a Haar-like classifier based on the AdaBoost algorithm can be used combining with an RGB-D camera to finish the task. The main process of this method is that after the Haar-like classifier has effectively recognized the target human body, we use the method of color segmentation to further extract the contour of the target body, and then maps the contour to the depth map and perform weighted summing with the depth map. The experimental results show that this method has high accuracy and strong practicability.