Motion detection for multiple moving targets by using an omnidirectional camera

This paper proposes a new real-time omnidirectional vision system for tracking multiple targets in indoor environments. Since traditional cameras suffer from the problem of having limited visual fields, an omnidirectional camera is used in our system to obtain the 360° view images of the global scene. First, omnidirectional images are changed into the cylindrical panoramic images. Then an adaptive background subtraction method is utilized to segment the moving regions. Human bodies qua the targets are tracked and identified by using their color information. Experimental results show that the proposed system performs well in indoor, complex environments.