实时运动物体分割是实现智能视频监控和视频交通流量检测等视觉系统的基础。目前，影响运动检测的因素主要有两个方面，一是背景建模的准确性;二是运动阴影的干扰。提出了基于无偏卡尔曼滤波器(UKF)的背景提取和阴影检测方法，构建整体的运动物体检测框架。该方法通过对背景和阴影建模，分别从帧间差分和背景差分两个层次综合分析像素值的动态变化特性，并利用色彩和亮度变化特性检测出运动阴影，最后借助UKF对两个模型参数进行在线更新，实现实时的运动物体分割。与现有算法相比，该方法背景跟踪速度快、运动检测效果好、计算量较小。实验结果证明了该方法的有效性和实用性。

Real-time segmentation of moving objects in image sequences is a fundamental step in many vision systems including intelligent visual surveillance and visual traffic analysis.Moving object detection has two key problems presently，one is the accuracy of background modeling，and the other is the disturbance from moving shadows.This paper presents a real-time background subtraction and moving shadow detection method based on Unscented Kalman Filter(UKF)，and constructs the whole frame for moving object detection.Background and moving shadow are modeled firstly.Then，dynamic character of pixel value is analyzed through frame to frame differencing and background differencing，and moving shadows are detected using luminance and chromatic cues of shadow.Finally，model parameters are updated online using UKF，and real-time segmentation of moving objects are completed.Our method exceeds existing ones at background maintenance speed，moving detection effect，and it has simple computational procedure.The experiments demonstrate the effectiveness and practicality of this method.