Independent Moving Target Detection for Aerial Video Surveillance

摘要

Due to complexities faced in the task of independent moving target detection from an airborne moving platform, simple methods like image stabilization followed by temporal differencing or change detection cannot be properly used unless distracting factors like 3D parallax motions, ego-motion compensation errors and image noise have been well disposed of. Instead of trying efforts to reduce these distractions, we proposed an attribute based salient extrema detection algorithm to extract the most salient extrema as targets from the residual motion image. Without the need for image filtering, the independent moving targets can be efficiently detected by valuating extrema saliency with respect to different morphological attributes. Experimental results show the effectiveness and flexibility of our algorithm.