Real-time human detection using relational depth similarity features

Abstract

Many conventional human detection methods use features based on gradients, such as histograms of oriented gradients (HOG), but human occlusions and complex backgrounds make accurate human detection difficult. Furthermore, real-time processing also presents problems because the use of raster scanning while varying the window scale comes at a high computational cost. To overcome these problems, we propose a method for detecting humans by Relational Depth Similarity Features(RDSF) based on depth information obtained from a TOF camera. Our method calculates the features derived from a similarity of depth histograms that represent the relationship between two local regions. During the process of detection, by using raster scanning in a 3D space, a considerable increase in speed is achieved. In addition, we perform highly accurate classification by considering of occlusion regions. Our method achieved a detection rate of 95.3% with a false positive rate of 1.0%. It also had a 11.5% higher performance than the conventional method, and our detection system can run in real-time (10 fps). © 2011 Springer-Verlag Berlin Heidelberg.