基于视频图像识别的铁路雪情分析及预测方法研究

发布时间：2018-05-11 16:59

本文选题：视频图像识别 + 风吹雪　；参考：《北京交通大学》2016年硕士论文

【摘要】：随着我国经济的增长,新建高速铁路的里程也在不断增加,交通事故也随之频发,因此交通安全保障技术越来越成为关注的焦点和研究的热点。雪灾是导致交通事故频发的一个重要因素,会引起铁路供电中断、线路受阻,严重影响了铁路的运输能力和运输安全。由雪灾造成的损失极其巨大,为了能将雪灾的损失降到最低,建立雪灾防护系统显得尤为重要,必须对降雪的信息有一个准确且及时的把握,比如当前雪的平均厚度以及单位时间内的降雪量等,从而实现更好的铁路行车预警。目前,我国铁路的测量主要应用人工测量和激光测量,其中激光测量比较准确而且反馈及时,现如今比较缺乏视频方面的分析雪情技术,遂本文针对雪情分析主要从以下几个方面进行研究：(1)通过图像处理技术对雪情进行分析,根据目前的运动目标检测技术,提出自己改进的基于实时视频图像识别的雪情检测分析方法。一共分为图像预处理、连续五帧法检测雪粒子、基于雪粒子轮廓尺寸的雪粒子提取、开窗选取有效区域、推算实时降雪强度五个部分,实时统计雪粒子密度,通过统计分级进而估计监控点处的实时降雪强度。(2)研究了风吹雪的形成机理,归纳了雪粒子的运动形式主要有蠕移、跃移和悬移,跃移占了绝大部分。风吹雪的堆积主要取决于风速场的大小,之后简单分析了风吹雪的堆积形式。之后分析建立铁路降雪量堆积模型,以铁路降雪量堆积模型作为预测分析积雪的主预测模型,雪的物理性质、当前的降雪量、历史雪害信息等作为模型计算的修正条件,采用定性与定量相结合、计算机仿真与专家经验相结合的方法进行雪情预测分析。最后提出相应的防护措施等。
[Abstract]:With the economic growth of our country, the mileage of new high-speed railway is also increasing, and the traffic accidents are also frequent. Therefore, traffic safety and security technology has become the focus of attention and research focus. Snow disaster is an important factor leading to frequent traffic accidents, which will lead to the interruption of railway power supply and the obstruction of railway lines, which seriously affects the transportation capacity and transportation safety of the railway. The losses caused by snow disasters are extremely huge. In order to minimize the losses caused by snow disasters, it is particularly important to establish a snow disaster protection system. We must have an accurate and timely grasp of snowfall information. For example, the average thickness of the current snow and the amount of snow per unit time, so as to achieve better railway warning. At present, manual and laser measurements are mainly used in railway survey in our country, among which laser measurement is accurate and feedback is timely. Nowadays, there is a relatively lack of video analysis technology for snow condition. Therefore, this paper mainly studies the snow situation analysis from the following aspects: 1) through the image processing technology to analyze the snow situation, according to the current moving target detection technology, An improved snow detection and analysis method based on real-time video image recognition is proposed. It is divided into five parts: image preprocessing, detection of snow particles by five successive frames, extraction of snow particles based on the size of snow particles contour, selection of effective areas by window opening, calculation of the real time snowfall intensity, and real-time statistics of snow particle density. The formation mechanism of wind-blown snow is studied by statistical classification and estimation of the real-time snowfall intensity at the monitoring point. The main motion forms of snow particles are compacted movement, jump and suspension, which account for most of the snow particles. The accumulation of wind-blown snow mainly depends on the size of wind field, and then the accumulation form of wind-blown snow is simply analyzed. Then the railway snowfall accumulation model is established, and the railway snowfall accumulation model is used as the main prediction model, the physical properties of snow, the current snowfall amount, the historical snow damage information and so on as the correction conditions for the model calculation. The method of combining qualitative and quantitative analysis, computer simulation and expert experience is used to predict snow condition. Finally, the corresponding protective measures are put forward.
【学位授予单位】：北京交通大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：U216.412

【相似文献】