摘要
低能见度天气条件如大雾和低云对航空器的进近着陆安全产生很大威胁,可造成可控飞行撞地和跑道侵入两类事故,同时在低能见度天气下,机场容量大幅度降低,导致航班延误增加,影响航空公司营运效益。解决低能见度天气条件下航空器进近着陆问题对飞行安全和航空公司的营运有重大意义。
解决这一问题的核心是“可视化”,可通过多源图像配准融合技术来改善图像的质量从而提高可视化问题,目前多源图像配准融合技术自动化程度不高,运算速度满足不了实时处理的需求。随着达芬奇技术的不断发展,可利用达芬奇多通道视频处理平台DM6446-TVP5158对图像算法进行处理。
本文对达芬奇DM6446-TVP5158的软件和硬件结构进行深入分析和研究,基于达芬奇的ARM+DSP的双核架构,研究了多通道的视频切换、单通道连续帧及多通道视频加权求平均处理的方法。ARM端负责多通道切换,在DSP端进行视频加权求平均处理,处理的效果能满足实时处理要求。
为了能够提取跑道特征以运用于增强视觉系统中,基于集成开发平台CCS3.3,采用Hough变换进行跑道的特征提取,在特征提取之前做了二值化和边缘检测预处理工作,详细对比分析了各种边缘检测算法在达芬奇系统DSP上的运行性能,验证了边缘检测算法性能取决因素,主要有:算子本身复杂程度,阀值设定的大小以及算法本身在DSP上的优化等。得出了Hough变换是提取直线的有效手段的结论。
Low visibility weather such as fog and low cloud poses a threat to the aircraft’s security on the approach of landing, it will lead to two types of accidents such as controlled flight into terrain and runway intrusion. The airport capacity has a significant reduction in low visibility weather conditions, which causing the increase of the flight delay , consequently, it affects the airlines’s operational efficiency.
The core problem is how to realize visual image. The problem can be improved by the technology of multi-source image’s and registration and fusion .The automation of multi-source image’s registration and fusion is not high currently , thus ,it can’t meet the need of the video’s real-time processing. The DaVinci’s multi-channel video processing platform DM6446-TVP5158 is applied to the field of image algorithm calculation with the development of DaVinci technology.
The paper analysed the software and hardware structure of DaVinci DM6446-TVP5158, and researched the multi-channel video switching, the single-channel consecutive frames’weighted average processing and the multi-channel’s weighted average based on the DaVinci’s dual core ARM+DSP architecture. ARM was responsible for multi-channel video switching, the algorithm of video weighted average was finished on DSP. The result met the requirements of real-time video processing .
In order to extract the characteristics of runway and apply the characteristics to enhanced flight vision system, the paper used Hough transform to extract the characteristics of runway based on integrated development platform CCS3.3. The edge detection and image binarization were done before feature extraction, the various edge detection algorithms’s performance which running on the DaVinci DSP were compared detailedly. The paper validated the factors of edge detection algorithm’s performance .They were the complexity of the algorithm, the setting value of threshold and the algorithm’s optimization running on DSP. The results demonstrated that Hough transform is an effective means of extracting straight line.
引文
[1]朱代武,何光勤.目视和仪表飞行程序设计[M].成都:西南交通大学出版社,2004:189-199,196-198
[2]王睿.可见光与红外图像融合技术研究[D].长春:长春理工大学,2008
[3]刘卫光,郭师经,周利华.红外与可见光图像实时配准融合系统[J].红外技术,2004,26(5):66-68
[4]王蕾.图像配准技术及应用研究[D].西安:西安电子科技大学,2007
[5]郑小龙.数字视频应用将把产业推向另一高峰[J].电子产品世界,2006,(4):26-27
[6]方进.面向数字视频的达芬奇(DaVinci)技术[J].电子产品世界,2005,(21):143-144
[7]TI. DavinciTM技术概览[EB/OL], http://www.ti.com.cn/davinci, 2008-9
[8]王旭.基于Davinci技术的视频监控系统[D].吉林:吉林大学,2008
[9]秦云川.基于DM6446的视频监控开发平台的设计与实现[D].长沙:湖南大学,2008
[10]裴科.基于TMS320DM6446的嵌入式数字视频系统设计[D].太原:太原理工大学,2007
[11]叶志龙.基于TMS320DM6446的嵌入式视频系统设计[D].太原:太原理工大学,2007
[12]付玉.基于达芬奇平台的微光视频实时处理系统的研究与实现[D].北京:北京交通大学,2008
[13]李希国.基于达芬奇平台的微光视频处理系统关键技术的研究与实现[D].北京:北京交通大学,2009
[14]胡慧之.基于DSP的微光视频图像实时处理的系统软件设计[D].南京:南京理工大学,2004
[15]季娜.基于DSP的雾天监控视频图像清楚化处理方法研究[D].大连:大连海事大学,2009
[16]胡新华,邓元勇,王先平.基于达芬奇技术的日面活动图像处理与识别系统[J].计算机应用研究,2008,25(10):3165-3168
[17]陈曦.基于高速DSP的图像跟踪技术研究[D].南京:南京理工大学,2009
[18]李露曦.微光与红外视频融合系统的图像处理器研究[D].南京:南京理工大学,2009
[19]魏弋彭.基于达芬奇平台的图像融合算法的实现和加速[D].上海:上海交通大学, 2010
[20]TI. DM644x DaVinciTM Technology Workshop Student Guide[Z]. Technical Training,2007
[21]TI. TMS320DM6446 Digital Media System-on-Chip[Z]. SPRS283G,2010
[22]张起贵,张胜,张刚,等.最新DSP技术——“达芬奇”系统、框架和组件[M].北京:国防工业出版社,2009:1-2
[23]彭启琮.达芬奇技术——数字图像/视频信号处理新平台[M].北京:电子工业出版社,2008:42-43
[24]TI. TMS320DM644x DMSoc Video Processing Front End(VPFE)[Z]. SPRUE38H,2010
[25] TI. TMS320DM644x DMSoc Video Processing Back End(VPBE)[Z]. SPRUE37C,2008
[26]TI. TVP5158,TVP5157,TVP5156 Four-Channel NTSC/PAL Video Decoders Data Manual[Z]. SLES243C,2010
[27]赵勇,袁誉乐,丁锐. DAVINC技术原理与应用指南[M].南京:东南大学出版,2008:93-94,101-103,119-121,122-142
[28]TI. Codec Engine Application Developer User’s Guide. SPRUE67D,2007
[29]合众达电子. SEED-DTK6446实验手册[EB/OL],http://www.seeddesp.com ,2008-12
[30]TI. Codec Engine Server Integrator User’s Guide[z]. SPRUED5B,2007
[31]崔晶.如何构建达芬奇的DSP Server[J].中国公共安全,2007,(8):121-126
[32]崔晶,帮您快速入门TI的Codec Engine[EB/OL], http://focus.ti.com.cn/cn/general/docs/gencontent.tsp?contentId=61575,2007
[33]TI. xDAIS-DM(Digital Media) User Guide[Z]. SPRUEC8B,2007
[34]李怀琼,陈钱,隋修宝,等.多帧累加平均技术在红外实时图像处理中的应用[J].激光与红外,2005,35(12):978-982
[35]张晓勤,岳继光,宋志光.基于Linux环境下视频采集的实现[J].计算机与现代化,2004,(1):18-20
[36]余谦,刘任庆.基于达芬奇技术的视频采集系统研究[J].广播与电视技术,2008,(3):93-96
[37]Michael H Schimek,Bill Dirks,Hans Verkuil,et al. Video for Linux Two API Specification Revision 0.24[EB/OL]. http://v4l2spec.bytesex.org,2008
[38]董晨.基于Davinci平台的数字视频技术研究[D].北京:北京交通大学,2007
[39]TI. TMS320 DSP/BIOS User’s Guide[z]. SPRU423F,2004
[40]TI. EncodeDecode Demo for the DVEVM/DVSDK 1.2[Z]. SPRAAH0A,2007
[41]程世陵,徐世宁,严林芳.民用飞机驾驶舱综合显示系统的发展趋势[J].民用飞机设计与研究,2009,(4):1-7
[42]卢惠民,郑志强.一种新的用于检测直线的快速Hough变换[J].计算机应用,2005,25(10):2379-2380
[43]张铮,王艳平,薛桂香.数字图像处理与机器视觉——Visual C++与Matlab实现[M].北京:人民邮电出版社,2010:360-361
[44]高守传,姚领田. Visual C++——数字图像处理与工程应用篇[M].北京:中国铁道出版社,2006:91-92
[45]张宏林.精通Visual C++数字图像处理典型算法及实现[M].北京:人民邮电出版社,2008:327-329
[46]TI. TMS320C64x+ DSP Image/Video Processing Library(V2.0.1)Programmer’s Guide[Z]. SPRUF30A,2008
[47]TI. Code Composer Studio Development Tools V3.3 Getting Started Guide[z]. SPRU509H,2006