Stereoscopic Human Detection in a Natural Environment

The algorithm presented in this paper is designed to detect people in real-time from 3D footage for use in Augmented Reality applications. Techniques are discussed that hold potential for a detection system when combined with stereoscopic video capture using the extra depth included in the footage. This information allows for the production of a robust and reliable system. To utilise stereoscopic imagery, two separate images are analysed, combined and the human region detected and extracted. The greatest benefit of this system is the second image, which contains additional information to which conventional systems do not have access, such as the depth perception in the overlapping field of view from the cameras. We describe the motivation behind using 3D footage and the technical complexity of human detection. The system is analysed for both indoor and outdoor usage, when detecting human regions. The developed system has further uses in the field of motion capture, computer gaming and augmented reality. Novelty comes from the camera not being fixed to a single point. Instead, the camera is subject to six degrees of freedom (DOF). In addition, the algorithm is designed to be used as a first filter to extract feature points in input video frames faster than real-time.