中文摘要：

针对目前在车辆检测中广泛应用的混合高斯模型（GMM）存在的缺陷，提出了一种改进的GMM运动车辆检测方法。对于GMM运行过程中“鬼影”长期存在的缺陷，通过采用新的权值和方差更新方法，加速“鬼影”的消除，改善其车辆检测性能；对于传统的GMM对所有像素点均采用固定分布数建模造成的内存空间浪费，通过设定一个分布数上限值，对未达到上限值的像素点采用分布数自适应变化的方法，有效地减少模型总分布数，节约内存空间。实验结果表明，改进后的GMM在“鬼影”的消除和计算速度上具有较大的优势。

英文摘要：

To overcome the defects of the Gaussian mixture model （GMM） in widely used vehicle cletec- tion method, an improved algorithm for moving vehicle detection based on Gaussian mixture model （GMM） is proposed. For the defect of long-term ＂ghosts＂ in Gaussian mixture model, a new up＇late method of weights and variances is employed to accelerate the elimination of ＂ghosts＂, and the performance of vehicle detection can be improved. Additionally,in the existing GMM,all of the pixels are modeled by fixed number of distributions, so main memories are wasted. To save memories, a self-adapting method is adopted. For the pixels whose distribution numbers are not up to maximum,the approach of adaptive change for distribution numbers is used to effectively decrease the total number of distributions and save memory space. Experimental results show that the improved GMM method provides superior performance in the elimination of ＂ghosts＂ and computing speed for vehicle detection.