Background subtraction with a moving camera

Abstract

Moving object segmentation with a nonstationary camera is a difficult problem due to the motion of both camera and the object. A moving object segmentation method is proposed in this work to be used in pan-tilt-zoom (PTZ) cameras. The method is based on composing scene mosaic and applying Gaussian mixture background subtraction algorithm after constructing a background model using the mosaic. Background subtraction is performed by mapping the frames captured during camera's course of motion to the background mosaic. The proposed mosaic building method requires less number of picture correspondences when compared to known methods. The success of the proposed segmentation method is demonstrated by the conducted experiments.