基于帧差法的运动车辆阴影去除算法  

Algorithm of moving vehicle shadow elimination based on frame difference

作　　者：李萌[1] 郑娟毅[1] 门瑜 

机构地区：[1]西安邮电大学通信与信息工程学院,陕西西安710121

出　　处：《电视技术》2016年第10期11-14,共4页Video Engineering

基　　金：国家自然科学基金项目(11401469);陕西省自然科学基金研究计划项目(2014JQ1030)

摘　　要：在视频交通车辆目标检测中,阴影问题是影响其检测准确性的关键问题之一。为了解决这个问题,提出了一种结合单模高斯模型和帧差法的运动目标阴影去除方法。首先通过单模高斯模型背景建模获取前景包括阴影在内的目标图像,再结合帧差法去除阴影。实验结果证明,该方法得到的车辆目标比较完整,并较好地去除了运动车辆阴影。Transport vehicles in the video object detection, shadow problem is one of the key issues affecting the detection accuracy. To solve this problem, a combination of moving objects shadow single-mode Gaussian model and frame difference method of remov- al methods is proposed. Firstly, get the target image includes shadows prospects, including through the single-mode Gaussian model, background modeling, combined with the frame difference remove the shadow. Experimental results show that the vehicle target obtained by this method is more complete, and can get better shadow addition to moving vehicles.

关 键 词：运动阴影去除 帧间差分法 单模高斯模型 车辆检测 

分 类 号：TN391[电子电信—物理电子学]