基于改进高斯混合模型的实时运动目标检测与跟踪*

为提高运动目标检测与跟踪的可靠性，提出了一种基于改进高斯混合模型的实时运动目标检测与跟踪算法。该算法建立可自动调节分布数目的高斯混合背景模型，通过背景减除获取前景图像；利用目标相邻帧的连续性分割运动目标；在此基础上将传统的颜色直方图模型进行改进，提高目标颜色分布的可信度，进而根据目标的位置、大小和颜色构造运动目标全局匹配相似度函数，实时完成运动目标检测与跟踪。利用大量的监控视频数据进行验证，结果表明，与传统的检测跟踪算法相比，该算法减少了计算量，提高了复杂背景情况下运动目标检测与跟踪的可靠性。

Moving object real-time detection and tracking based on improved Gaussian mixture model

To improve the reliability of moving objects detection and tracking, this paper presented a moving object real-time detection and tracking algorithm based on improved Gaussian mixture model background modeling. The algorithm established Gaussian mixture background model which could adjust the number of distribution automatically and detected moving objects by background subtraction, separated the moving object by the continuity of adjacent frames. On this basis, enhanced the reliability of objective color distribution by improving traditional color histogram model. Then achieved real-time moving object detection and tracking by the similarity function which was constructed by the position, size and color of objects. Lots of experiments based on a large amount of practical monitor video data were completed, and the results show that comparing with the traditional algorithm, the new algorithm reduces the computational complexity and improves the reliability of moving objects detection and tracking in complex environment.