An Effective Method to Detect Moving Objects under PTZ Camera

Due to the changes of background, detection of moving objects under PTZ camera becomes critical and full of challenges. To solve this problem, a new method based on K-means clustering is proposed in this paper. Generally speaking, the motion vector of background is not equal to the motion vector of moving object, so that the moving target is extracted from the background. Gray projection method is used to extract Region of Interest (ROI). The global motion vector and motion vectors of blocks in ROI are obtained by block-matchin g method. K-means clustering is adopted to segment the moving target. By this way moving targets are detected accurately. Experimental results show that the method is not only suitable for fast-moving target, but can also extract the moving target under dynamic background quickly and precisely. So it is indicated that the method has some engineering value.