论文部分内容阅读
本文提出一种固定背景下的运动目标自动分割技术．首先，提出了一种新的运动变化检测方案．然后利用递归高阶统计的方法从高斯噪声中提取运动变化．同传统的高阶统计方法相比，递归高阶统计由于利用前ｎ 帧的图像信息，所以能够更有效地抑制噪声，检测小目标，并且分割效果较好．
This paper presents a fixed background automatic moving target segmentation technology. First of all, a new motion change detection scheme is proposed. The recursive high-order statistics are then used to extract the motion variation from Gaussian noise. Compared with the traditional high-order statistics, the recursive high-order statistics can effectively suppress the noise, detect the small target, and the segmentation effect is better because of using the image information of the first n frames.