基于改进的自适应混合高斯背景模型的车辆检测方法

针对静止摄像机条件下运动车辆的检测问题,提出一种改进的自适应混合高斯背景模型的方法.该方法初始时通过三帧差分法判断运动目标所在区域,运用提出的区域背景更新算法生成初始背景图像,然后在Stauffer等人提出的自适应混合高斯背景模型的基础上融入帧间差分和背景差分相结合的方法用于判定运动目标区域和背景区域,通过对背景区域和运动目标区域设置不同的学习率来更新背景模型,提高了模型的收敛速度.实验结果表明,同传统检测方法相比,改进的算法能较快地初始化背景模型并能有效地检测出运动车辆,有较强的鲁棒性和较好的自适应能力.

Based on Improved Adaptive Background Mixture Gaussian Model Vehicle Detection

An improved adaptive background mixture Gaussian model has been proposed under the conditions of static camera motion detection problem of vehicles.Initially,it determines the moving target region by three frame difference methods,updates the algorithm by regional background to generate the initial background image.Then on the basis of adaptive Gaussian mixture background model put forward by Stauffer,it integrates the frame difference and background difference methods to determine moving target zone and background zone.It sets different learning rates according to background zone and the moving target zone to update the background model and it raises the convergence speed.The experimental results show that,compared with traditional detection methods,the improved algorithm can more quickly initialize background model and efficiently detect moving vehicles and also it has strong robustness and good adaptability.