中文摘要：

提出一种简化的核密度估计模型KDE（KernelDensityEstimation）动态背景建模算法，构建了基于ARMll的视频动态背景分割系统。不仅能大幅度降低传统KDE背景建模的所需的运算量和存储量，提高处理速度，而且还具有良好的运动前景分割效果。系统完成了视频动态背景的建模、运动前景分割、视频显示、图片压缩保存等功能，具有较好的性价比。

英文摘要：

The paper presents a simplified KDE （Kernel Density Estimaiton） dynamic background modeling algorithm and constructs a dynamic background segmentation hardware system based on ARMll. It can significantly reduce the computation cost and memory storage required for the traditional KDE algorithm and improve the processing speed. It can also yield favorable mobile foreground segmentation results. The system completes dynamic background modeling, mobile foreground segmentation, video display, image compression and storage. It is cost effective.