基于中值滤波和多特征融合的粒子滤波车辆跟踪算法研究

PARTICLE FILTER ALGORITHM RESEARCH FOR VEHICLE TRACKING BASED ON MEDIAN FILTERING AND MULTI-FEATURE FUSION

为获得复杂环境下较好的车辆跟踪结果,提出了一种基于中值滤波和多特征融合的粒子滤波车辆跟踪算法.在粒子滤波跟踪框架中,首先将车辆图像从RGB(Red,Green,Blue)颜色空间转换到HSV(Hue,Saturation,Value)颜色空间,利用中值滤波对视频图像进行预处理后,再利用prewitt算子检测图像边缘信息,然后融合颜色和边缘两个特征,实现复杂环境下的车辆跟踪.实验结果表明,该算法具有较好的目标跟踪精度和鲁棒性,能有效解决单一颜色特征下容易导致目标车辆跟踪丢失的问题.

In order to obtain better vehicle tracking results in complex environment,particle filter algorithm research for vehicle tracking based on median filtering and multi-feature fusion was presented. In particle filter tracking framework,the vehicle images were transformed from RGB space to HSV space,and the images of the video were preprocessed by median filtering,then the edge information was detect by making use of prewitt algorithm. And the using multiple features fusion based on color and edge information,implement the vehicle tracking under complex background. The experimental results shows that this algorithm has better accuracy of object tracking and robustness,It can effectively solve the problem of target vehicle tracking loss under the condition of single color histogram.