Improved difference method for fast-moving object recognition in complex background
Xiaoping Li, Xulong Ren, Song Ye, Liu Qiongxin · 2014
Aiming at the problem that it is difficult to identify the fast moving objects in complex background, the paper proposes a method that establishes the inter-frame background sequence, and then multiply the pixel grayscale in the difference image. It defines constraint rules for public background image, gets moving target region with higher-order cumulant and block grayscale, and further highlights the edges of moving objects, largely eliminating the natural environment noise from lighting, scene and shooting angle changes. Experimental results show that this method can maintain real-time tracking of moving objects, identify non-overlapping moving targets, and provide the basis for subsequent target analysis.