基于目标识别与跟踪的嵌入式铁路异物侵限检测算法研究

Study on Railway Embedded Detection Algorithm for Railway Intrusion Based on Object Recognition and Tracking

铁路线路异物侵限是威胁行车安全的一个重要隐患。基于机器视觉与嵌入式技术设计了异物侵限自动检测系统,利用FPGA和ARM芯片实现了图像采集处理硬件平台。提出异物目标分类和运动行为分析相结合的嵌入式异物侵限检测算法。算法采用两级判别结构,首先利用支持向量机及一组特征向量对背景差分图像得到的异物目标进行分类,根据分类结果滤除大部分行进列车目标,之后运用Kalman滤波器设计目标跟踪算法,对其余目标进行行为和运动趋势分析,滤除其中非侵限干扰信息提高报警准确率,并对有侵限趋势的异物提前预警。实验表明,该系统能够有效地识别检测区域内的异物目标,系统侵限报警准确率达到97.11%,平均检测频率达13帧/s。

Railway obstacle intrusion is a great potential threat to the safety of train operation. An automatic in- trusion detection system is designed based on machine vision and embedded technology, in which the hardware platform for image acquisition and processing is realized with FPGA and ARM micro-processor. An embedded algorithm of intrusion obstacle detection is proposed combining obstacle classification with moving behavior a- nalysis. A two-stage discrimination structure is adopted in the algorithm. Firstly, the objects acquired from background subtraction images are classified with Support Vector Machine and a group of eigenvectors, and the most of the moving train targets are removed from the detection results. Subsequently, a Kalman-filter track- ing algorithm for the remaining objects is employed, to analyze the behavior and moving trend of the objects,. to filter out the non-intrusion interference information in order to improve the alarm accuracy, and to issue an early-warning on the obstacles with intrusion trend. The experiment results show that the system can detect the obstacles within the detection area effectively, achieving the warning accuracy rate of 97.11~ and the aver- age detection frequency of 13 frames/s.