【作者】 冯桂兰；

【导师】 田维坚；

【作者基本信息】 中国科学院研究生院（西安光学精密机械研究所） ， 光学工程， 2007， 博士

【摘要】 在夜间，为了车辆能够隐蔽行驶，需要保证无灯光条件下可以实时地获得前方路况和方向信息。本文针对这个需求提出并研制了新型的车载夜视导航系统。该系统是一套辅助有自主式无源导航功能的小型化宽视场微光夜视设备。本文的主要研究内容及其创新点表现在以下几个方面：(1)新型车载夜视导航系统的研究：由于现有夜视仪的视场角大约在30°到40°之间，并不能满足驾驶车辆的要求。为解决这个问题，本文提出新型的车载夜视导航系统并对其进行研究。已研制完成的车载夜视导航系统可以自动实时地提供宽视场微光视频图像，且辅助有隐蔽导航功能。(2)高速DSP图像拼接系统的研究：该系统主要研究如何实现宽视场微光视频图像。通过对多种现有扩大视场方法进行分析，本文采用图像拼接技术扩大观察范围。并考虑实际工程应用要求，选用数字信号处理器作为主处理器。系统可以脱离计算机运行，保证具有小型化和便于携带的特点。实验结果表明高速DSP图像拼接系统可以提供无缝的宽视场微光夜视图像。系统以20帧／秒输出视场角为120°的宽视场微光视频图像，系统的后视模块为车辆倒车提供了保障。(3)高速DSP图像拼接系统硬件设计研究：通过对高速DSP和存储器等技术进行研究，论证设计高速DSP图像拼接系统总体框架，对使用的器件进行合理选择，设计系统完整的硬件电路原理图，并完成系统全部电路板的技术设计与制作。通过系统的初步调试表明系统的硬件设计是正确可行的，为宽视场微光视频图像的实现奠定了硬件平台基础。(4)高速DSP图像拼接系统图像拼接算法及其软件设计的研究：对多种图像拼接算法进行分析比较，综合数字信号处理器的特点，针对实际需要提出“基于亚象素相位相关的图像配准算法+基于小波变换的图像融合算法”的系统图像处理总体算法模式。通过对高速DSP图像拼接系统需要完成的功能模块进行分析，采用基于DSP／BIOS框架进行软件设计，并合理利用DSP的EDMA和EMIF资源，实现图像拼接系统的多线程软件设计。实验结果中测得视频图像的速度约为20帧／秒，即0.05秒刷新一帧，按照车辆在夜间最大的行驶速度为每小时50公里，即每秒行驶14米，得到汽车行驶0.7米图像就可以刷新一帧，因此视频图像的速度完全达到车辆行驶的要求，证明系统的图像处理总体算法模式和软件设计合理可用。(5)数字磁罗经系统的研究：该系统的研究为驾驶员在黑暗中无法把握方向提供解决途径。通过对现有的多种导航方式进行研究比较，并考虑隐蔽性的要求，本文提出新型的数字磁罗经系统。该系统使用被动方式测量方位信息，因此克服卫星定位导航系统易受干扰和受制于人的缺点。通过分析基于地磁的导航方式带来的自差和磁差产生原理，结合泊松方程，提出自差的消除算法，改进传统磁罗经不能自动消除自差和磁差的缺点。数字磁罗经原理样机能够隐蔽提供当前车辆行驶的方向信息，且成功实现了自差和磁差的自动消除，将精度保证在0.5°以内。更多还原

【Abstract】 In order to drive snugly the vehicle at night, the forward situation and directionof vehicle must be acquired snugly and real timely without the assistance ofillumination. The new Night Vision and Navigation System on Vehicle (NONSV) hasbeen presented and studied. It is a minitype wide field low-light-level night visiondevice which has independent and passive navigation function as assistantance.The main researched contents and achievements are as follows:(1) A elementary prototype of the new NONSV is studied and produced. Becausethe Field of View (FOV) of the existing night vision device is about 30°to 40°, thedemand of driving vehicles can’t be satisfied. In order to solve this problem, NONSVhas been presented and studied in the dissertation. The designed NONSV canautomatically and real timely provide a wide field forward road image and have thecovert navigation function.(2) The study of Image Mosaics System based on Digital Signal Processor(IMS-DSP). The system was mainly studied on how to acquire wide fieldlow-light-level images. The image mosaics technique had been adopted in the systemby analysising many methods to enlarge the FOV. According to the demand of theactual project, DSPs was adopted as main processor. The system can run withoutcomputer so that it ensures miniaturization and carry-home trait. The test result provesIMS-DSP can obtain a seamless wide field low-light-level video image at 20FPS withFOV of 120°and its back vision module provides a substitute of the backup light.(3) The study of the hardware design of IMS-DSP. The overall framework ofIMS-DSP was demonstrated and designed by studying the high speed DSPs andmemory technology. The used device was chosen reasonably, the whole hardwarecircuit plot was designed, and the PCB was designed and made. By the primarydebugging for the system, it is proved that the hardware design of the system iscorrect and feasible, and that the hardware platform is established to realize the capture of wide field low-light-level video images.(4) The study of image mosaicing algorithm. According to DSPs characteristic,to meet the actual needing, many image mosaicing algorithms were analysised andcompared, and the entire image processing algorithm mode of the system wasdesigned by combining the image registration algorithm based on subpixel phasecorrelation with image fusion algorithm based on the wavelet transformation. Byanalysing the function module of IMS-DSP, the software design methods onDSP/BIOS was adopted, EDMA and EMIF resources were utilized reasonably, andmultithreading software design of the system was obtained. The speed of video imagewas 20FPS in the test result, the frame was refurbished in 0.05 second, the vehicle’smaximum speed was 50 kph (14 meters per second), and the vehicle ran 0.7 meterswhen the image was refurbished, so the speed of video image can satisfy the practicalneed. It is proved that the entire image processing algorithm mode of the system andsoftware design method was reasonable.(5) The study of digital compass system. The study of the system provides anapproach to solve that the driver can’t obtain direction at night. By studying andcomparing many navigation techniques, and considering the need of concealment, thenew digital compass system was studied and produced. The passive mode was used tomeasure direction information in the system, so the shortcoming of the satelliteposition navigation system that is prone to be disturbed and controlled by others canbe overcomed. By analyzing the forming principle of deviation and magneticdeviation, and using Poisson equation, an algorithm to remove deviation wasestablished, and the new proposed algorithm can overcome the shortcoming of thetraditional compass that can’t automatically remove the deviation and magneticdeviation. The elementary protype of the digital compass system can snugly providethe direction information, and can successfully and automatically remove thedeviation and magnetic deviation, and its precision is 0.5°.更多还原