基于道路背景特征的车辆阴影消除算法研究-软件工程专业论文.docxVIP

摘 要 随着中国机动车市场大规模的发展、以及道路拥堵和机动车及驾驶员之间突 出矛盾。我觉得智能交通在中国，特别是在大中型城市是一个非常有前景的研究 和实践领域。 目前国内外针对智能交通系统进行了大量的研究工作。智能交通最重要的一 部分就是针对车辆的识别，其它的系统开发和系统应用全部是在车辆和车牌识别 的基础上完成的。为了实现系统级的应用我们要在拍摄的图像中进行车辆及其运 动轨迹的识别（拍车辆是为车牌信息、运动轨迹是违法信息）这样针对车辆的跟 踪就非常重要了，国内外的研究课题也相对较多。目前由于车辆的阴影存在对计 算机的跟踪处理有很大的不良后果和影响,所以如何有效地消除运动车辆的阴影 是本篇论文主要研究的内容。 本篇论文就是要解决这个问题，我要对现有的图像裁剪阴影的算法进行优化 工作，在计算机进行图像分析的同时应用本算法快速的裁剪车辆所形成的阴影， 精确定位场景中运动的对象，以便得到正确的车辆信息。 本方法充分利用阴影部分图像亮度虽然变化，但颜色饱和度基本不变化这一 特性，将运动图像分割成若干小块图像，比对前、后两帧视图中对应小块图像的 亮度和颜色饱和度的变化情况，确定非影子区域，从而精确定位实际运动物体。 从实验结果上面分析此算法的效果非常明显，既大大降低了系统的运算压 力，又显著提高了系统的识别效率，车辆识别准确，为其它应用程序的开发提供 了非常稳定、真实的数据来源。 关键词：智能交通 交通违法 车辆识别 阴影 ABSTRACT As Chinas vehicle market expands in a large scale, and the conflicts between traffic jam , vehicles and drivers stand out, I believe smart traffic in cities in China, especially medium and large ones, is a promising research and a practical project. The author has made substantial research in the national and international smart traffic system. The most important part of a smart traffic system is the identification of vehicles. The development and the application of other systems all rely on the identification of vehicles and license plate. In order to put the system into practice, we must make identification of vehicles and their movement locus on the image we have shot ( Taking a photo of a vehicle is to identify and match any information concerning the vehicle , its movement locus in case any illegal traffic behavior is committed). Therefore, to keep a track of vehicles becomes critically important. So far, there have been numerous research subjects home and abroad in this aspect. At present, the existence of vehicle shadows exerts rather negative impact on computerized tracking. Therefore, how to eliminate the shadows of vehicles become the theme of the paper. This essay will solve this issue, the optimizing work concerning the calculation method of the present image cutting shadow is to be carried out, while the computer is analyzing the imag