非高斯噪声背景下计算机视觉目标跟踪方法

针对杂波背景下计算机视觉目标跟踪问题,提出一种非高斯噪声背景下计算机视觉目标跟踪方法.在视频目标运动模型和观测模型的基础上引入了柯西混合噪声模型,对非高斯噪声运动目标的状态进行建模;然后,在传统高斯噪声粒子滤波的框架内给出文中方法的具体实现步骤.针对大面积遮挡和夜晚光照改变的极端情况下对路上行驶的车辆进行实时跟踪实验,结果表明:文中方法明显提升极端杂波环境下的目标运动过程的建模精度,有效提升目标跟踪精度.

Computer Vision Target Tracking Method UnderNon-Gauss Noise Background

Aiming at the problem of computer vision target tracking in clutter background, a computer vision target tracking method under non Gauss noise background is proposed. Based on the vision target moving model and observation model, the Cauchy mixed noise model is introduced to model the non Gauss noise moving object, and the non Gauss noise moving target state is modeled. The proposed method concrete implementation steps is realized in the framework of the traditional Gauss noise particle filter. For driving large area occlusion and night illumination change under extreme conditions on the road of the vehicle real-time tracking, the experimental results show that this method significantly improve the modeling accuracy of extreme clutter target motion process, effectively improve the accuracy of target tracking.