基于车载雷达和光学视频的车辆目标距离速度高精度估计方法

发布时间：2024-04-14 06:42

　　车载雷达可全天时全天候工作,是自动驾驶、智能无人系统的重要传感器,然而车载雷达的距离和速度测量精度较低,成为制约其应用效能的技术瓶颈。针对该问题,本论文分别从设计性能更优的信号处理及参数估计算法,以及融合雷达和光学视频测量结果两个方面开展研究,具体工作包括:(1)考虑OFDM信号体制,通过分析OFDM多载波回波模型,提出了改进的FFT算法和最大似然估计算法,通过充分利用多载频测量结果提高速度测量精度,并进?步给出了距离和速度估计的Cramer-Rao下界;(2)针对光学视频图像,研究了基于CNN模型的运动目标检测和位置速度估计算法,并在此基础之上研究了视频图像测量结果与雷达测量结果的关联,为融合处理提升测量精度奠定基础。论文方法经过仿真和MR3003雷达实际同步采集的雷达和视频数据进行了验证。

【文章页数】：80 页

【学位级别】：硕士

【文章目录】：
ABSTRACT
摘要
Abbreviations
Chapter 1:Introduction
    1.1 Motivation
    1.2 Background and Literature Review
    1.3 Thesis Organization
Chapter 2:Vehicle Radar
    2.1 Radar Signals
        2.1.1 Constant Frequency Pulses
        2.1.2 Linear Frequency Modulation Pulses
        2.1.3 FMCW Radar Signal
    2.2 FMCW Vehicle Radar(MR-3003-RD)
        2.2.1 Key Features
        2.2.2 Range neighbor delta(PRND)
        2.2.3 Speed neighbor delta(PSND)
    2.3 Application of MR3003-RD
    2.4 Problem Statement
    2.5 Summary
Chapter 3:Estimation methods for accurate Velocity Estimation
    3.1 OFDM Signal
    3.2 Conventional Algorithm for Parameter Estimation
        3.2.1 Range Resolution
        3.2.2 Doppler Resolution
    3.3 Improved Doppler estimation by Extended FFT method
        3.3.1 Symbol Constraints
    3.4 The real Data Experimentations with MR3003 radar
    3.5 Doppler Estimation by Maximum Likelihood Estimation
        3.5.1 Cramer Rao Lower Bound
    3.6 Summary
Chapter 4:Association of Radar Measurements and Video Tracking
    4.1 Vehicle Detection
        4.1.1 Data Set
        4.1.2 Convolutional Neural Network(CNN)
    4.2 Association of Radar Measurements and Optical Measurements
    4.3 Summary
Conclusion
    Future Work and Recommendations
References
Acknowledgement
Publications



本文编号：3954277