中文摘要：

针对运动目标检测中的背景重建问题，提出一种用 Kalman 滤波理论改进混合高斯背景模型的建模方法，进行背景图像的重建和更新。把当前帧目标图像与背景图像进行差分运算，检测出运动目标。通过红外图像库中标准数据集的测试，实验结果验证了该方法应用于红外图像运动目标检测的有效性。

英文摘要：

Background image reconstruction is an interesting research fields in motion obj ect detection.A method based on Kalman filter theory for Gauss mixture modeling improved about background image re-construction is presented in this paper.Motion target can be detected using background subtraction algo-rithm based on above it in surveillance video.Experimental results show the proposed method can be ef-fectively applied to detect infrared images moving obj ect.