基于视频图像的交通路口背景提取方法研究-物理电子学专业论文.docxVIP

摘要 I PAGE PAGE IV 摘要 利用视频图像对背景进行提取的方法作为一项基础而关键的研究，在各种领 域中发挥了重要的作用，其中一种应用是对前景目标进行检测，即对提取出的背 景图像，利用背景差方法提取与检测相应的前景目标。在交通领域中，可利用背 景差法提取交通路口场景中车辆，行人等前景运动目标交通流信息。这些交通流 信息在交通路口智能化的调度，检测车辆违规驾驶等方面有着重要的作用。所以 如何获取一个精确的交通路口场景背景是一项重要的工作。 针对上述的问题，本文分析了现有的背景提取算法如基于均值法的背景模型， 像素直方图统计背景模型，混合高斯背景模型等优劣，设计了一种基于多模态直 方图与空间时间相关性分类背景更新的改进算法对交通路口背景进行拟合和更 新，通过基于 EmguCV 机器视觉图像库的软件平台对采集到的视频图像进行仿真 来验证算法的有效性。 通过实验证明，本文采用的背景拟合方法更新算法，能够更好的对交通路口 的道路背景进行拟合，有着很好的稳定性，可以应对多模式背景，运动目标速度 缓慢，运动目标密集时造成虚假背景的情况。 关键词：智能交通 背景提取 多模态背景 交通路口 Ab Abstract III PAGE PAGE IV Abstract The use of video images for background extraction method as a basic and key research, play an important role in various fields, by extracting the background image, background difference method is utilized to extract and the prospect of the corresponding target detection. Especially in the field of transportation, can use background difference method to extract the vehicle in the traffic intersection, the pedestrian traffic flow moving foreground objects,etc. This plays an important role in intelligent traffic intersection of scheduling, inspection vehicle violation driving. So how to obtain an accurate scene background in traffic intersection is an important job. According to the above problem, this paper analyzes the existing algorithms for background extraction background model, such as average background model,histogram statistical background model, and gaussian mixture background model, this paper proposes a time and time correlation classification based on multimode and histogram background update algorithm to fitting and update of background traffic intersection, through EmguCV based machine vision image library software platform for video images were obtained by simulation to verify the effectiveness for the algorithm. Experiments show that the proposed background update algorithm can better fitting and updating the road traffic intersection background, also has a good stability and can deal with multimodal backgro