Vehicle Tracking and Speed Measure

Hikvision's Line Haul Management SolutionBoost productivity and minimize transportation risksLogistics operators and drivers face numerous hassles for the goods they transport and the unpredictable transportation process. This makes transportation security paramount to their business.For logistics operators, the primary concerns center on getting timely, accurate, and comprehensive status updates of their freight trucks. They are seeking answers to dispatch monitor and schedule the trucks precisely, to ensure the trucks arrive safely, to examine the containers in real-time.Now, Hikvision figures out the real challenges and offers the optimal solution. Read on to find out more.BackgroundHikvision’s SolutionsCommon ChallengesAbsent of Data or Evidence of AccidentsDifficult to determine accident causes and trace back the evidenceof traffic accidents Vehicle Tracking & MonitoringSupport real-time GPS tracking and post-event video evidence trace back with HD video securityCostly Operation with At-Risk DrivingCarry more driving risks due to long working hours, physical burdens, harsh weather, etc.ZDriver Protection & AssistanceHigh Operation Costs Due to Loss and TheftFuel lossGoods loss due to spoilage, rejected loads, etc.Property Management & ProtectionEstablish extensive IOT applications with assorted sensors, including door status sensors, temperature sensors, fuel level sensor, and mobile recorders.Performance & Operation AnalysisSupport holistic management to improve decision making efficiency with assorted reports from HikCentral ProfessionalInefficient Operations ManagementDifficult to optimize operation Isolated data cannot accurately reflect the state of operationsProvide audio communication and alarm for in-time respondAchieve comprehensive protection with the intelligent driving assistance• Forward Collision Warning (FCW) • Lane Departure Warning (LDW) • Pedestrian Collision Warning (PCW) • Blind Spot Detection (BSD)• Headway Monitoring Warning (HMW) •Driver Safety Management (DSM)Solution FunctionsVideo SecuritySafe and stable transportation are always the priority for logistic operators. However, the driving process can be tough to manage, especially when determining the cause of an accident and exonerate innocent drivers. The lack of data or evidence can be a serious impediment. What can be done?Hikvision’s Line Haul Management Solution provides real-time monitoring to support drivers and continually records video to provide evidence of incidents. Operators can browse alarm records or timestamps.A 360° panoramic camera with all-around views supports safe driving. Drivers can improve awareness of surroundings and operate vehicles more safely with the all-around views when turning or moving in reverse. In emergencies, they can notify the platform using a convenient alarm button or by speaking directly through a two-way intercom . Additionally, driving alarms for events such as deviations and speeding go straight to the platform with real-time video and vehicle location to maintain safe operations.Deviation/Fence-crossing Incidents Panic stop Sharp turnAlarm triggerAuto pops up the incident messageHanding the alarmReportAlarm check and settlement Send to admins ForensicsEmergency Anomaly detection Overspeed FCW LDW PCWHMWBSDADASManual alarmCollision alarm Anomaly detection alarmAlarm ManagementReal-time Tracking HD MonitoringMoving vehicleStopped vehicle Vehicle in alarm O ine vehicleGoods ManagementDamage, vandalism, and theft of goods and merchandise are always the uncontrollable threats for operators, because they cannot examine those goods during the transportation process. Here, Hikvision’s IoT Module can help to enhance goods management with extensive IoT applications and assorted sensors, consisting of door status sensors, temperature sensors, fuel level sensor, and mobile recorders .*Door Status Sensors and Temperature Sensors support third-party sensors and are ready for project deployment. Door Status Sensor needs no integration. For Temperature Sensor, DVR customization is needed.Intelligent Driving AssistanceDifficulties like long hours, physical burdens, harsh weather, and more, make driving safely a challenge. And at-risk driving is very dangerous for merchandise, drivers, and general road safety. Our intelligent driving assistance system offers intelligent driver protection and assistance. It features Forward Collision Warning (FCW) , Lane Departure Warning (LDW) , Pedestrian Collision Warning (PCW) , Blind Spot Detection (BSD) , Headway Monitoring Warning (HMW) and Driver Safety Management (DSM). Drivers can enjoy a safe driving with intelligent analysis server to help identify potential dangers and trigger audio warnings.Moreover, drivers can navigate prudently with optimal awareness with the abnormal driving behavior detection from DSM camera.Door Open 16: 50 PMDoor Open 18: 13 PMDoor Status RecordsEventsT his advanced module provides temperature and fuel consumption monitoring and integrity assurance of merchandise. An operator can inspect the rear door status throughout the journey to make sure that all goods are intact during transit. For cold chain services, temperature sensors can help shipping professionals check container temperatures to maintain product quality , with temperature history and real-time alarms for abnormalities. Also, the IOT module can identify a sudden drop in fuel level, and analyze usage for maximum efficiency .Command Center - HikCentral ProfessionalOperators have difficulty evaluating their drivers’ performance, because isolated data cannot accurately reflect the real state of operations. However, HikCentral Professional can centralize management with helpful functions like video monitoring, recording, playing back footage, real-time positioning, data & statistics, and more. A variety of reports present operators with great visibility across every aspect, helping operators handle emergencies instantly and make smarter decision.*Note: Units of measure vary across regions. Hikvision applies the appropriate units in any given application/installation.Number of events per 100 km8612Event & Alarm StatisticsLiter per 100 kilometers202119LLLFuel Consumption StatisticsPunctual delivery989597Arrival StatisticsMileage StatisticsMiles covered this month9,800km11,50013,200kmkmOnline Hour StatisticsHours worked this month181205234hourshourshoursDriver A Driver B Driver C%number of eventsnumber of eventsnumber of events%%Solution StructureCommand CenterHikCentralMonitorIntercom Mobile Camera Panic Button 360° Surround View Camera360° Surround View Monitor HostVideo SecurityVideo, Alarm & IntercomRS232Mobile RecorderBSD Camera DSM Camera ADAS Camera Door status IO RS485RS485Fuel LevelTemperature RecordIntelligent Driving AssistanceVideo-based Intelligent Analysis Video, ADAS & AlarmGoods ManagementSolution Extension with Assorted SensorsFor large-scale trucks, they carry more risks due to the great inertia and more blind zones. Hikvision provides com-prehensive solution to guarantee the safe transportation, including video security, intelligent driving assistance, and goods management, etc.Front Camera 360° Surround View System Rear CameraDoor Closed Sensor Temperature SensorDVRBSD Camera IntercomSide Camera Indoor Camera ADAS CameraLCD PanelDBA CameraFuel Level Sensor Panic ButtonBox trucks or lorries and single-chassis vehicles360° Surround View SystemCommon Applications360° Surround View Host 5LCD Panel6360° Surround View Camera 14~View of cam 4Cam 1Cam 3View of cam 2256413DSMSide CameraDVRFuel Level SensorBSD Camera Indoor CameraADAS CameraLCD PanelDBA CameraPanic ButtonIntercom2For big rigs without power supply in container, the solution canfocus on the video security and intelligent driving assistancepart to assure the safety of driving and transportation.Big rigs and varioustrucks with trailersVideo Recording & RetrievalFence Crossing AlarmReal-time MonitoringRoute & Speed ControlDSMProduct FeaturesHD Video Security•All-round cameras with high definition, including 720p and 1080p, can provide better recording, tracking, and deterrenceMobile CameraMobile DVRsReliable Supply, Modular Design• Power supply supports 8-36 VDC• Electromagnetic immunity design conforms to the automobile ISO-7637 electromagnetic standard •Suitable for all types of vehiclesAnti-Seismic Design• Top-rated shockproof hard drives installed •Patented metal air bag damping technologyUninterrupted Power Design• First to be utilized in the industry•Ensures video is written to the hard drive at the moment of power failure5sPerfect Video Recording• Supports a variety of storage media• Supports SD card redundancy recording when if hard drive is damaged•Supports fire box storage / redundancy recordingSDGPS Filtering•Equipped with high-sensitivity antenna and GPS filtering algorithms, GPS Filtering resolves problems with fixed-point drift, velocity drift, and other abnormal data. Moreover, regarding data transmission, information will be maintained when offline or Internet disconnection; transmission will be restored when back online.Aviation Gyroscope•Built-in G-sensor & gyroscope sensor provide emergency response to collision and rollover accidents to restrict or eliminate dangerous driving habits such as rapid acceleration, deceleration or braking, and turning too sharplyDual System Backup•Dual system firmware is built in the vehicle host to ensure normal operation of the device after restartingIP68 Waterproof Rating• Water pressure: 400 KPa • Water flow: 100L/min • Depth: 2.5 – 4 m •Test duration: >3 mins*Only outdoor analog camera achieves IP68.AE-VC163T-ITS AE-VC263T-ITS• Front and Rear Dual-Channel Vehicle-Mounted Camera • 720p / 1080p @ 30 fps• Min. illumination: 0.1 Lux @ (F1.2, AGC ON) , 0 Lux with IR • ICR for rear camera and full color for front camera • Built-in microphone• Rear Camera: 2.1 mm; Field of View: H: 125°, V: 80°, Diag: 148°• Front Camera: 2.1 mm; Field of View: H: 125°, V: 66°, Diag: 154°• Adjustment Angle: Tilt: 0 - 360°•CE / FCC / E-markAE-VC159T-S AE-VC259T-S• Front Vehicle-Mounted Camera • 720p / 1080p @ 30 fps• Min. illumination: 0.01 Lux @ (F1.2, AGC ON) • Supports full color• Wide-angle lens, 2.1 mm, Field of View: H: 127°, V: 73°• SNR: 42 dB• Consumption: ≤ 2.2 W • CE / FCC / E-mark AE-VA136T• Front Vehicle-Mounted Camera • 720p @ 30 fps• Min. illumination: 0.01 Lux @ (F1.2, AGC ON) • Supports full color• Wide-angle lens, 2.1 mm, H FOV 127°, V FOV 73°• SNR: 42 dB• Consumption: ≤ 1 W •CE / FCC / E-mark• 720p / 1080p @ 30 fps • Metal Case• 0.1 Lux @ (F1.2, AGC ON)• Operating temperature: -40 - 75° C • Ingress protection: IP68• Adjustment Angle: Tilt: 0°to 90°• Supports auto day & night switch • Supports right & left monitor mode •SNR: 62 dBAE-VC153T-IT AE-VC253T-ITWorks with AI DVR to achieve BSDAE-VC154T-IT• 6 mm lens • 720p @ 30 fps• Min. illumination: 0.1 Lux @ (F1.2, AGC ON) , 0 Lux with IR • Tilt: 0 - 50°, Pan: 0 - 5°, Rotate: 0 - 360°• 2 x 940 nm IR • 3 m IR distance • SNR: 42 dB• Consumption: ≤ 2.5 W Works with AI DVR to achieve DSMAE-VC155T• Windshield camera • 6 mm lens • 720p @ 30 fps• Min. illumination: 0.1 Lux @ (F1.2, AGC ON) • Tilt: 0 - 65°• WDR: 120 dB • SNR: 42 dB•Consumption: ≤ 2 WWorks with AI DVR to achieve ADASHikvision Colombia****************************Hikvision CzechT +420 29 6182640*********************Hikvision Europe T +31 23 5542770 **********************Hikvision Egypt T +20223066117**********************Hikvision Azerbaijan T +994 50 369 81 57*****************************Hikvision BrazilT +55-11-3318-0050***************************Hikvision Canada T +1-866-200-6690**************************Hikvision Australia T +61-2-8599-4233*********************Hikvision IndiaT +91-22-6855 9944************************Hikvision Indonesia T +6221 2933 9366*****************************Hikvision IsraelT +972 79 5555590**************************Hikvision ItalyT +39 0438 6902 *********************Hikvision FranceT +33(0)1 85 330 450 *********************Hikvision Hungary KFT*********************Hikvision Hong Kong , China **********************Hikvision Germany************************Hikvision New Zealand T 09 217 3127*********************Hikvision New Panama Sales.centralamerica @Hikvision Mexico T +52 55 2624 0110**************************Hikvision Pakistan T +92-2135147526************************Hikvision South Africa T +27 877018113*************************Hikvision Tashkent T +99-87-1238-9438 ****************Hikvision Singapore T +65 6684 4718 ****************Hikvision SpainT +34 91 737 16 55 *********************Hikvision Philippines************************Hikvision Russia T +7-495-669-67-99********************Hikvision Romania**************************Hikvision PolandT +48 22 460 01 50 *********************Hikvision Thailand****************************Hikvision Turkey T +90 216 521 70 70************************Hikvision UAET +971-4-4432090*********************Hikvision UK & Ireland T +44(0)1628 902 140 *********************Hikvision Uzbekistan T +998-71-233-55-50************************Hikvision USAT +1-909-895-0400***********************Hikvision Vietnam T +84 24 7300 7586*********************Hikvision Kenya**************************Hikvision Malaysia T +60327224000**********************Hikvision Korea T +82-1661-8138*************************Hikvision Kazakhstan T +7 (727) 291-75-88********************DS-1350HM• 1-ch build-in speaker and 1-ch build-in Mic • 12 VDC / 30 mA • 85 x 85 x 26 mm• Operating Temperature: -10 to 55° C • Wiring length: 6,000 mmDS-1530HMI• Built-in Gyroscope and Sensor • 5 VDC / 30 mA • 80 x 30 x 19 mm• Operating Temperature: -20 to 60° C • 100 ms sampling interval• 7-inch TFT-LCD, 800 x 480 RGB • 130° visual angle• Manual switch or Auto switch• 1-ch connect to the DVR, 1-ch connect to the rear camera • 2-ch alarm input • IP 54DS-MP1301DS-MP1302 (Touch Screen)Hikvision’s Line Haul Management Solution Boost productivity and minimize transportation risks。

车辆纵向速度估算算法现状及趋势专业：控制理论与控制工程班级：2008学生姓名：梁晋昌学号：20080201008导师：韩峻峰2010年3月8日车辆纵向速度估算算法现状及趋势梁晋昌（广西工学院电子信息与控制工程系，广西柳州545006）摘要：在车辆行驶过程中，纵向速度是车辆主动安全系统中的重要信息。

在制动防抱死(ABS)和驱动防滑系统 (ASR)中，纵向车速是计算纵向滑移率、保持车辆行驶稳定性的重要参数。

对现存的车辆纵向速度算法进行了分类综述，将其分为基于基本信息的直接计算方法和基于模型信息的间接计算方法两大类，对各种方法的优缺点进行了讨论，并对其发展趋势进行了展望。

关键词：纵向速度；速度估计；车辆模型Abstract: Vehicle is in motion the process, the vertical velocity of vehicle active safety systems in the important information. Anti-lock braking (ABS) and drive-slip system (ASR), the vertical speed is to calculate the vertical slip rate to maintain the stability of vehicles an important parameter. Vehicle longitudinal speed of the existing algorithms are classified overview of basic information will be divided into based on the direct calculation method and model-based information on the indirect method of calculating two categories, the advantages and disadvantages of various methods were discussed, and its development trends predicted.Key words: Longitudinal velocity; Velocity estimation; Vehicle Model0 引言在车辆行驶过程中，纵向速度是车辆主动安全系统中的重要信息。

第１３卷㊀第４期Ｖｏｌ．１３Ｎｏ．４㊀㊀智㊀能㊀计㊀算㊀机㊀与㊀应㊀用ＩｎｔｅｌｌｉｇｅｎｔＣｏｍｐｕｔｅｒａｎｄＡｐｐｌｉｃａｔｉｏｎｓ㊀㊀２０２３年４月㊀Ａｐｒ．２０２３㊀㊀㊀㊀㊀㊀文章编号：２０９５－２１６３（２０２３）０４－０１４７－０５中图分类号：ＴＰ３８９．１文献标志码：Ａ基于双目视觉的车辆检测跟踪与测距郭鹏宇（上海工程技术大学机械与汽车工程学院，上海２０１６２０）摘㊀要：由于道路上存在各种不安全因素，车辆检测跟踪并测距是自动驾驶技术的重要组成部分㊂本文将ＹＯＬＯｖ４－ｔｉｎｙ作为检测器使之加快模型检测速度且更适合在车辆嵌入式设备中使用㊂考虑到目标检测失败的情况，本文在历史缓冲区中存储以前的跟踪细节（唯一ＩＤ）和每个对象的相应标签，提出了一个基于中值的标签估计方案（ＭＬＰ），使用存储在前一帧的历史标签的中值来预测当前帧中对象的检测标签，使得跟踪错误最小化，并用双目摄像头获取图像检测车辆距离㊂测试新网络结构后，检测精度（ＭｅａｎＡｖｅｒａｇｅＰｒｅｃｉｓｉｏｎ，ｍＡＰ）为８０．１４％，检测速度较ＹＯＬＯｖ４相比提高了１８４％，检测到的距离误差平均在０．５％左右㊂关键词：ＹＯＬＯｖ４－ｔｉｎｙ；目标跟踪；中值算法；双目测距Ｖｅｈｉｃｌｅｄｅｔｅｃｔｉｏｎ，ｔｒａｃｋｉｎｇａｎｄｒａｎｇｉｎｇｂａｓｅｄｏｎｂｉｎｏｃｕｌａｒｖｉｓｉｏｎＧＵＯＰｅｎｇｙｕ（ＳｃｈｏｏｌｏｆＭｅｃｈａｎｉｃａｌａｎｄＡｕｔｏｍｏｔｉｖｅＥｎｇｉｎｅｅｒｉｎｇ，ＳｈａｎｇｈａｉＵｎｉｖｅｒｓｉｔｙｏｆＥｎｇｉｎｅｅｒｉｎｇＳｃｉｅｎｃｅ，Ｓｈａｎｇｈａｉ２０１６２０，Ｃｈｉｎａ）ʌＡｂｓｔｒａｃｔɔＤｕｅｔｏｖａｒｉｏｕｓｕｎｓａｆｅｆａｃｔｏｒｓｏｎｔｈｅｒｏａｄ，ｖｅｈｉｃｌｅｄｅｔｅｃｔｉｏｎ，ｔｒａｃｋｉｎｇａｎｄｒａｎｇｉｎｇａｒｅｔｈｅｉｍｐｏｒｔａｎｔｐａｒｔｏｆａｕｔｏｍａｔｉｃｄｒｉｖｉｎｇｔｅｃｈｎｏｌｏｇｙ．Ｉｎｔｈｉｓｐａｐｅｒ，ＹＯＬＯｖ４－ｔｉｎｙｉｓｕｓｅｄａｓａｄｅｔｅｃｔｏｒｔｏｓｐｅｅｄｕｐｍｏｄｅｌｄｅｔｅｃｔｉｏｎａｎｄｉｓｍｏｒｅｓｕｉｔａｂｌｅｆｏｒｖｅｈｉｃｌｅｅｍｂｅｄｄｅｄｄｅｖｉｃｅｓ．Ｃｏｎｓｉｄｅｒｉｎｇｔｈｅｆａｉｌｕｒｅｏｆｏｂｊｅｃｔｄｅｔｅｃｔｉｏｎ，ｔｈｉｓｐａｐｅｒｓｔｏｒｅｓｔｈｅｐｒｅｖｉｏｕｓｔｒａｃｋｉｎｇｄｅｔａｉｌｓ（ｕｎｉｑｕｅＩＤ）ａｎｄｔｈｅｃｏｒｒｅｓｐｏｎｄｉｎｇｌａｂｅｌｏｆｅａｃｈｏｂｊｅｃｔｉｎｔｈｅｈｉｓｔｏｒｙｂｕｆｆｅｒ，ａｎｄｐｒｏｐｏｓｅｓａｍｅｄｉｕｍ－ｂａｓｅｄｌａｂｅｌｅｓｔｉｍａｔｉｏｎｓｃｈｅｍｅ（ＭＬＰ），ｗｈｉｃｈｕｓｅｓｔｈｅｍｅｄｉａｎｖａｌｕｅｏｆｔｈｅｈｉｓｔｏｒｙｌａｂｅｌｓｔｏｒｅｄｉｎｔｈｅｐｒｅｖｉｏｕｓｆｒａｍｅｔｏｐｒｅｄｉｃｔｔｈｅｄｅｔｅｃｔｉｏｎｌａｂｅｌｏｆｔｈｅｏｂｊｅｃｔｉｎｔｈｅｃｕｒｒｅｎｔｆｒａｍｅ，ｓｏｔｈａｔｔｒａｃｋｉｎｇｅｒｒｏｒｓａｒｅｍｉｎｉｍｉｚｅｄ．Ｔｈｅｉｍａｇｅｓｏｂｔａｉｎｅｄｂｙｂｉｎｏｃｕｌａｒｃａｍｅｒａａｒｅｕｓｅｄｔｏｄｅｔｅｃｔｖｅｈｉｃｌｅｄｉｓｔａｎｃｅ．Ａｆｔｅｒｔｅｓｔｉｎｇｔｈｅｎｅｗｎｅｔｗｏｒｋｓｔｒｕｃｔｕｒｅ，ｔｈｅｄｅｔｅｃｔｉｏｎａｃｃｕｒａｃｙ（ＭｅａｎＡｖｅｒａｇｅＰｒｅｃｉｓｉｏｎ，ｍＡＰ）ｉｓ８０．１４％，ｔｈｅｄｅｔｅｃｔｉｏｎｓｐｅｅｄｉｓ１８４％ｈｉｇｈｅｒｔｈａｎｔｈａｔｏｆＹＯＬＯｖ４，ａｎｄｔｈｅｄｅｔｅｃｔｅｄｄｉｓｔａｎｃｅｅｒｒｏｒｉｓａｂｏｕｔ０．５％ｏｎａｖｅｒａｇｅ．ʌＫｅｙｗｏｒｄｓɔＹＯＬＯｖ４－ｔｉｎｙ；ｔａｒｇｅｔｔｒａｃｋｉｎｇ；ｍｅｄｉａｎａｌｇｏｒｉｔｈｍ；ｂｉｎｏｃｕｌａｒｄｉｓｔａｎｃｅｍｅａｓｕｒｅｍｅｎｔ作者简介：郭鹏宇（１９９５－），女，硕士研究生，主要研究方向：智能网联汽车㊂收稿日期：２０２２－０５－２４０㊀引㊀言在自动驾驶辅助系统中，基于传感器，采用车辆检测㊁跟踪㊁测距等一系列计算机视觉算法进行环境感知，辅助系统就能得到车辆周围信息，以保证驾驶员安全驾驶㊂基于视觉的车辆检测及测距系统主要应用在道路交通场景下，用于辅助检测前方目标以及进行距离预警，其性能好坏主要依赖于采用的车辆检测算法㊂目前，在使用相机进行目标检测时，多采用传统的机器视觉检测方法㊂对于前方车辆目标，该方法首先根据车辆的局部特征，如阴影㊁边缘纹理㊁颜色分布等特征生成感兴趣区域；然后利用对称特征等整体特征对感兴趣区域进行验证㊂在从产生感兴趣区域到验证感兴趣区域的过程中，为了达到实时检测的要求，一般需要对图像进行灰度化，并对灰度化后的图像进行阴影分割和边缘分析㊂因此，对于相机获得的图像，传统的机器视觉的车辆检测方法通常找到感兴趣区域的车辆的特点和梯度直方图特征（ＨＯＧ［１］），ＳＩＦＴ［２］特征或Ｈａａｒ－ｌｉｋｅ［３］特征通常用于获得前面的假设检验区域车辆，即ＲＯＩ区域；此后用这些特征训练ＳＶＭ［４］或Ａｄａｂｏｏｓｔ［５］车辆检测分类器，计算车辆图像的特征值，并根据车辆特征值的大小与前方车辆进行判断，得到前车的假设测试区域验证，完成对前车的检测㊂而上述传统的机器视觉检测方法本质上是通过人工选择特征进行识别和分类㊂在复杂场景中，人工特征的数量会呈几何级数增长，这对前面车辆的识别率也有很大的影响㊂这种方法更适合在某种特定场景下的车辆识别，因为其数据规模并不大，泛化能力则较差，很难实现快速和准确的复杂应用场景的检测㊂近年来，随着卷积神经网络（ＣＮＮ）的应用，出现了许多算法㊂一阶段方法包括ＳＳＤ［６］㊁ＹＯＬＯ系列［７－８］㊁ＲｅｔｉｎａＮｅｔ［９］㊂两阶段方法包括ＦａｓｔＲ－ＣＮＮ［１０］和ＦａｓｔｅｒＲ－ＣＮＮ［１１］㊂最近提出的最先进的ＹＯＬＯ－ｖ４［１２］具有很高的检测精度和检测速度㊂目前，对于多目标车辆轨迹跟踪技术主要可分为两大类㊂一类是传统方法，如利用背景差分法㊁帧差法㊁光流法等方法提取运动目标，传统方法部署方便，资源消耗低，但受先验知识限制，跟踪稳定性差，准确性不高㊂另一类是基于卷积神经网络的㊁称为深度学习的方法，深度学习方法可以学习更多的目标特征，能在连续的视频帧中检测出目标对象㊂深度学习方法精度高，但其计算量较大，实时性不高，因此，基于视频跟踪的车辆检测算法仍需改进㊂研究可知，基于视觉相机的测距方法主要有单目测距和双目测距两种㊂这２种方法的共同特点是通过相机采集图像数据，随后从图像数据中得到距离信息㊂单目检测方法的优点是成本低，缺点是对检测精度的依赖过大㊂此外，从实用的角度来看，在汽车上安装单目相机时，由于汽车的颠簸，汽车的俯仰角经常发生变化，导致精度显著下降㊂双目测距的方法是通过计算２幅图像的视差直接测量距离㊂１㊀车辆检测与跟踪本文使用的目标检测算法是ＹＯＬＯｖ４－ｔｉｎｙ，其中ＹＯＬＯ表示ＹｏｕＯｎｌｙＬｏｏｋＯｎｃｅ，由Ｂｏｃｈｋｏｖｓｋｉｙ等学者开发㊂ＹＯＬＯｖ４－ｔｉｎｙ是ＹＯＬＯｖ４的压缩版本，虽在平均精度方面受到了影响，但却可以在低计算能力下高效运行㊂与未压缩版本的４个ＹＯＬＯ头相比，ＹＯＬＯｖ４－ｔｉｎｙ只使用了２个ＹＯＬＯ头，并使用了２９个预训练卷积层作为基础㊂ＹＯＬＯ各变量参数设置见表１，卷积层各变量参数设置见表２㊂㊀㊀上一代ＹＯＬＯ的非ｍａｘｐｏｏｌ抑制（ＮＭＳ）等遗留特性和一些新特性㊁包括加权剩余连接（ＷＲＣ）㊁Ｍｏｓａｉｃ数据增强在内有效提高了算法在模糊图像中识别类的能力，降低了识别类所需的处理能力㊂ＹＯＬＯｖ４－ｔｉｎｙ提供了较高的帧率，同时具有中间地带平均精度与常用模型并列㊂在本文中，使用ＹＯＬＯｖ４－ｔｉｎｙ算法作为车辆的检测器，并且使用ＤｅｅｐＳＯＲＴ［１３］算法作为初始车辆跟踪器㊂表１㊀ＹＯＬＯ各变量参数设置Ｔａｂ．１㊀ＹＯＬＯｐａｒａｍｅｔｅｒｓｅｔｔｉｎｇｓＹＯＬＯ变量参数Ｍａｓｋ０，１，２ａｎｃｈｏｒｓ１０，１４，㊀２３，２７㊀３７，５８㊀８１，８２㊀１３５，１６９㊀３４４，３１９ｃｌａｓｓｅｓ４ｎｕｍ６ｊｉｔｔｅｒ０．３ｓｃａｌｅ＿ｘ＿ｙ１．０５ｃｌｓ＿ｎｏｒｍａｌｉｚｅｒ１．０ｉｏｕ＿ｌｏｓｓｃｉｏｕｉｇｎｏｒｅ＿ｔｈｒｅｓｈ０．７ｔｒｕｔｈ＿ｔｈｒｅｓｈ１ｒａｎｄｏｍ０Ｒｅｓｉｚｅ１．５ｎｍｓ＿ｋｉｎｄｇｒｅｅｄｙｎｍｓｂｅｔａ＿ｎｍｓ０．６表２㊀卷积层各变量参数设置Ｔａｂ．２㊀Ｔｈｅｃｏｎｖｏｌｕｔｉｏｎｌａｙｅｒｐａｒａｍｅｔｅｒｓｅｔｔｉｎｇｓ卷积层变量参数ｂａｔｃｈ＿ｎｏｒｍａｌｉｚｅ１ｆｉｌｔｅｒｓ６４ｓｉｚｅ３ｓｔｒｉｄｅ２ｐａｄ１ａｃｔｉｖａｔｉｏｎｌｅａｋｙ㊀㊀图１显示了２个ＩＤ及其前３个标签㊂对于ＩＤ＃１３７的车辆，本文方法预测的标签用加黑来标记㊂[1255,739,1421,856][960,719,1006,758]车辆I D 137[1255,739,1421,859][955，721，1006，758][1255,739,1421,856][952，722，1006，758]目标检测标签1图１㊀应用ＭＬＰ后的历史缓冲区示例图Ｆｉｇ．１㊀ＡｈｉｓｔｏｒｙｂｕｆｆｅｒｅｘａｍｐｌｅａｆｔｅｒａｐｐｌｙｉｎｇＭＬＰ８４１智㊀能㊀计㊀算㊀机㊀与㊀应㊀用㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀第１３卷㊀㊀㊀本文使用历史缓冲区来调整和改进每个检测标签的视觉质量和在帧中的显示㊂如果有任何车辆检测标签缺失，那么本文的ＭＬＰ为该车辆生产一个估计的检测标签㊂延时使用一系列的检测标签前存储在历史缓冲区来预测未检测到车辆的检测标签ＩＤ在给定的框架（见图１）㊂条件估计为特定车辆检测标签，标签ＩＤ应该至少在２个连续帧出现㊂为了预测缺失的检测标签，本文对当前帧ｔ应用以下公式：Ñ＝ｌ（ｔ－１）（ｉ）－ｌ（ｔ－２）（ｉ）（１）ｌ（ｔ）（ｉ）＝ｌ（ｔ－１）（ｉ）＋Ñ（２）㊀㊀这里，ｌｘｍｉｎ，ｙｍｉｎ，ｘｍａｘ，ｙｍａｘ()表示每个车辆ＩＤ基于调整边界框标签的中值，Ñ表示边界框位置的变化从时间戳（ｔ－２）到（ｔ－１）；ｉ表示每辆车唯一的ＩＤ㊂２㊀双目测距双目视差示意如图２所示㊂由图２可知，２个摄像头的中心距是Ｂ，两个摄像头观察同一点Ｐ，Ｐ１的坐标为（Ｘ１，Ｙ１），Ｐ２的坐标为（Ｘ２，Ｙ２），由于２台相机始终是平行的，高度相同，所以在左右２张图像中Ｐ点的Ｙ坐标是相同的，在Ｘ方向上存在视差㊂因此，可以通过标定相机内外参数来确定图像坐标与世界坐标之间的关系㊂双目视差原理如图３所示㊂PZ 2X 2P 2Z 1P 1O 2Y 2Y 1BC 2X 1O 1C 1x 1y 1y 2x 2图２㊀双目视差示意图Ｆｉｇ．２㊀ＳｃｈｅｍａｔｉｃｄｉａｇｒａｍｏｆｂｉｎｏｃｕｌａｒｐａｒａｌｌａｘdfB aP RC RA RbP LX LX RPC LA L图３㊀双目视差原理图Ｆｉｇ．３㊀Ｐｒｉｎｃｉｐｌｅｏｆｂｉｎｏｃｕｌａｒｐａｒａｌｌａｘ㊀㊀图３中，ＣＬ和ＣＲ是２个摄像头的光学中心，摄像头之间的距离是ｂ，相机的焦距为ｆ，Ｐ点在左右图像的投影点为ＰｌＸｌ，Ｙｌ()，ＰｒＸｒ，Ｙｒ()，ＡＬ，ＰＬ＝ＸＬ，ＡＲ，ＰＲ＝ＸＲ，ＰＲ，Ｂ＝ａ，从三角形相似关系可知：ｄ－ｆｄ＝ａａ＋ｘＲｄ－ｆｄ＝ｂ－ｘＬ＋ｘＲ＋ａｂ＋ｘＲ＋ａ（３）㊀㊀由式（３）可知：ａ＝ｂｘＲｘＬ－ｘＲ－ｘＲ（４）㊀㊀由此，空间中Ｐ点到相机的距离为：ｄ＝ｆａ＋ｘＲｘＲ＝ｂｆｘＬ－ｘＲ（５）㊀㊀Ｐ在相机坐标系中的三维坐标可以由几何关系得到：Ｘ＝ｂｘＬｘＬ－ｘＲＹ＝ｂｙｘＬ－ｘＲＺ＝ｂｆᶄｘＬ－ｘＭìîíïïïïïïïï（６）㊀㊀对于车辆的测距，本文取检测到的边界框内每辆车的中心来表示被检测物体到双目相机中心的距离㊂３㊀实验结果与分析将ＹＯＬＯｖ４－ｔｉｎｙ与其他常用的目标检测算法进行比较，将其ｍＡＰ与ＦＰＳ进行比较，得到表３中的结果㊂本文提出的车辆检测与跟踪方法使用了ＴｅｎｓｏｒＦｌｏｗ库和基于ＹＯＬＯｖ４－ｔｉｎｙ模型的ＤｅｅｐＳＯＲＴ算法㊂经综合比较，使用ＹＯＬＯｖ４－ｔｉｎｙ的精度和检测速度是可以接受的，精度比ＹＯＬＯｖ３－ｔｉｎｙ高，速度比ＹＯＬＯｖ４的方法更快㊂ＹＯＬＯｖ４－ｔｉｎｙ模型检测车辆效果如图４所示㊂表３㊀各模型帧率和ｍＡＰ对比Ｔａｂ．３㊀ＦｒａｍｅｒａｔｅａｎｄｍＡＰｃｏｍｐａｒｉｓｏｎ模型ｍＡＰ／％帧率（ＦＰＳ）ＹＯＬＯｖ４８５．０８１４．１２ＹＯＬＯｖ４－ｔｉｎｙ８０．１４４０．１１ＹＯＬＯｖ３８３．３２１６．９９ＹＯＬＯｖ３－ｔｉｎｙ６９．０３５２．７７９４１第４期郭鹏宇：基于双目视觉的车辆检测跟踪与测距图４㊀ＹＯＬＯｖ４－ｔｉｎｙ模型检测车辆效果Ｆｉｇ．４㊀ＣａｒｓｖｉｄｅｏｄｅｔｅｃｔｉｏｎｕｓｉｎｇＹＯＬＯｖ４－ｔｉｎｙｍｏｄｅｌ㊀㊀使用本文方法前后汽车的标签变化曲线如图５所示㊂对于ＩＤ＃３９的车辆，图５（ａ）是使用方法前，图５（ｂ）是使用本文方法后，相同的汽车标签变得更加平滑㊂X （b o u n d i n g b o x 的中心）100200300400500600700800730720710700690680670Y （b o u n d i n g b o x 的中心）（ａ）使用本文方法前100200300400500600700800730720710700690680670X （b o u n d i n g b o x 的中心）Y （b o u n d i n g b o x 的中心）（ｂ）使用本文方法后图５㊀使用本文方法前后汽车的标签变化Ｆｉｇ．５㊀Ｔｈｅｌａｂｅｌｃｈａｎｇｅｓｂｅｆｏｒｅａｎｄａｆｔｅｒｕｓｉｎｇｔｈｅｍｅｔｈｏｄｉｎｔｈｉｓｐａｐｅｒ㊀㊀在目标跟踪时，从历史缓冲区中预测缺失标签的方法往往会产生不好的结果，因为对象检测器的可视化结果经常显示不稳定和闪烁的边框标签㊂在应用本文的基于中值的算法后，可以得到高度稳定的标签㊂因此，本文方法提高了目标检测器的视觉性能，并为目标检测器和跟踪器提供了对缺失标签的更好估计㊂利用双目相机取检测到的边界框内每辆车的中心来表示被检测物体到双目相机中心的距离㊂仿真测试结果见表４㊂从距离测试的结果来看，测试精度相对较高，基本保持在０．５％ ０．６％之间㊂表４㊀测量结果分析Ｔａｂ．４㊀Ｔｈｅｍｅａｓｕｒｅｄｒｅｓｕｌｔｓａｎａｌｙｓｉｓ实验组数测量距离／ｃｍ实际距离／ｃｍ误差／％１１５６７．００１５５９．１１０．５０３５２１６５５．００１６４６．１４０．５３５３３１７３８．００１７２９．１６０．５０８６４１８９３．００１８８３．１７０．５１９３５１９８３．００１９７１．２００．５９５１６２２３６．００２２２３．２２０．５７１６７２４８９．００２４７５．２６０．５５２０４㊀结束语本文介绍了一种用于自动驾驶的实时检测跟踪与测距系统㊂通过本文提出的实时同步方法，该系统方便了车辆实时同步检测；利用双目摄像头，ＹＯＬＯｖ４－ｔｉｎｙ和ＤｅｅｐＳＯＲＴ算法对车辆进行检测和跟踪，并提出中值标签预测方法优化跟踪效果，同时实现了对前方车辆的精确测距㊂整个系统在检测和测距方面取得了较高的精度和实时性㊂对于自动驾驶的应用，该系统可以结合许多智能技术，如目标预警㊁自动避障等㊂与此同时，该系统还有很大的改进空间㊂在检测方面，通过优化算法提高检测性能，通过训练更多类型的物体，如行人㊁非机动车等，为自动驾驶提供更多的道路信息㊂在这个系统中，测距是指从双目相机的中心到物体的距离㊂在实际情况下，车辆的具体位置到物体的距离可以根据相机的安装位置和车辆的实际长度来计算㊂通过优化双目测距算法，可以提高测距精度㊂参考文献［１］ＴＡＩＧＭＡＮＹ，ＹＡＮＧＭｉｎｇ，ＲＡＮＺＡＴＯＭＡ，ｅｔａｌ．ＤｅｅｐＦａｃｅ：Ｃｌｏｓｉｎｇｔｈｅｇａｐｔｏｈｕｍａｎ－ｌｅｖｅｌｐｅｒｆｏｒｍａｎｃｅｉｎｆａｃｅｖｅｒｉｆｉｃａｔｉｏｎ［Ｃ］／／２０１４ＩＥＥＥＣｏｎｆｅｒｅｎｃｅｏｎＣｏｍｐｕｔｅｒＶｉｓｉｏｎａｎｄＰａｔｔｅｒｎＲｅｃｏｇｎｉｔｉｏｎ．Ｃｏｌｕｍｂｕｓ，ＯＨ，ＵＳＡ：ＩＥＥＥ，２０１４：１７０１－１７０８．［２］ＭＡＸｉａｏｘｕ，ＧＲＩＭＳＯＮＷＥＬ．Ｇｒｉｍｓｏｎ．Ｅｄｇｅ－ｂａｓｅｄｒｉｃｈｒｅｐｒｅｓｅｎｔａｔｉｏｎｆｏｒｖｅｈｉｃｌｅｃｌａｓｓｉｆｉｃａｔｉｏｎ［Ｃ］／／ＴｅｎｔｈＩＥＥＥＩｎｔｅｒｎａｔｉｏｎａｌＣｏｎｆｅｒｅｎｃｅｏｎＣｏｍｐｕｔｅｒＶｉｓｉｏｎ．Ｂｅｉｊｉｎｇ：ＩＥＥＥ，２００５：１１８５－１１９２．［３］ＸＵＱｉｎｇ，ＧＡＯＦｅｎｇ，ＸＵＧｕｏｙａｎ．Ａｎａｌｇｏｒｉｔｈｍｆｏｒｆｒｏｎｔ－ｖｅｈｉｃｌｅｄｅｔｅｃｔｉｏｎｂａｓｅｄｏｎＨａａｒ－ｌｉｋｅｆｅａｔｕｒｅ［Ｊ］．ＡｕｔｏｍｏｔｉｖｅＥｎｇｉｎｅｅｒｉｎｇ，２０１３，３５（４）：３８１－３８４．（下转第１５７页）０５１智㊀能㊀计㊀算㊀机㊀与㊀应㊀用㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀㊀第１３卷㊀参考文献［１］ＳＨＩＪｉａｎｐｉｎｇ，ＴＡＯＸｉｎ，ＸＵＬｉ，ｅｔａｌ．Ｂｒｅａｋａｍｅｓｒｏｏｍｉｌｌｕｓｉｏｎ：ｄｅｐｔｈｆｒｏｍｇｅｎｅｒａｌｓｉｎｇｌｅｉｍａｇｅｓ［Ｊ］．ＡＣＭＴｒａｎｓａｃｔｉｏｎｓｏｎＧｒａｐｈｉｃｓ（ＴＯＧ），２０１５，３４（６）：１－１１．［２］ＹＡＮＧＤｏｎｇ，ＱＩＮＳｈｉｙｉｎ，Ｒｅｓｔｏｒａｔｉｏｎｏｆｄｅｇｒａｄｅｄｉｍａｇｅｗｉｔｈｐａｒｔｉａｌｂｌｕｒｒｅｄｒｅｇｉｏｎｓｂａｓｅｄｏｎｂｌｕｒｄｅｔｅｃｔｉｏｎａｎｄｃｌａｓｓｉｆｉｃａｔｉｏｎ［Ｃ］／／ＩＥＥＥＩｎｔｅｒｎａｔｉｏｎａｌＣｏｎｆｅｒｅｎｃｅｏｎＭｅｃｈａｔｒｏｎｉｃｓａｎｄＡｕｔｏｍａｔｉｏｎ．Ｂｅｉｊｉｎｇ，Ｃｈｉｎａ：ＩＥＥＥ，２０１５：２４１４–２４１９．［３］ＡＢＢＡＴＥＡ，ＡＲＥＮＡＲ，ＡＢＯＵＺＡＫＩＮ，ｅｔａｌ．Ｈｅａｒｔｆａｉｌｕｒｅｗｉｔｈｐｒｅｓｅｒｖｅｄｅｊｅｃｔｉｏｎｆｒａｃｔｉｏｎ：ｒｅｆｏｃｕｓｉｎｇｏｎｄｉａｓｔｏｌｅ［Ｊ］．ＩｎｔｅｒｎａｔｉｏｎａｌＪｏｕｒｎａｌｏｆＣａｒｄｉｏｌｏｇｙ，２０１５，１７９：４３０－４４０．［４］ＬＹＵＷ，ＬＵＷｅｉ，ＭＡＭｉｎｇ．Ｎｏ－ｒｅｆｅｒｅｎｃｅｑｕａｌｉｔｙｍｅｔｒｉｃｆｏｒｃｏｎｔｒａｓｔ－ｄｉｓｔｏｒｔｅｄｉｍａｇｅｂａｓｅｄｏｎｇｒａｄｉｅｎｔｄｏｍａｉｎａｎｄＨＳＶｓｐａｃｅ［Ｊ］．ＪｏｕｒｎａｌｏｆＶｉｓｕａｌＣｏｍｍｕｎｉｃａｔｉｏｎａｎｄＩｍａｇｅＲｅｐｒｅｓｅｎｔａｔｉｏｎ，２０２０，６９：１０２７９７．［５］ＹＩＸｉｎ，ＥＲＡＭＩＡＮＭ．ＬＢＰ－ｂａｓｅｄｓｅｇｍｅｎｔａｔｉｏｎｏｆｄｅｆｏｃｕｓｂｌｕｒ［Ｊ］．ＩＥＥＥＴｒａｎｓａｃｔｉｏｎｓｏｎＩｍａｇｅＰｒｏｃｅｓｓｉｎｇ，２０１６，２５（４）：１６２６－１６３８．［６］ＧＯＬＥＳＴＡＮＥＨＳＡ，ＫＡＲＡＭＬＪ．Ｓｐａｔｉａｌｌｙ－ｖａｒｙｉｎｇｂｌｕｒｄｅｔｅｃｔｉｏｎｂａｓｅｄｏｎｍｕｌｔｉｓｃａｌｅｆｕｓｅｄａｎｄｓｏｒｔｅｄｔｒａｎｓｆｏｒｍｃｏｅｆｆｉｃｉｅｎｔｓｏｆｇｒａｄｉｅｎｔｍａｇｎｉｔｕｄｅｓ［Ｃ］／／ＩＥＥＥＣｏｎｆｅｒｅｎｃｅｏｎＣｏｍｐｕｔｅｒＶｉｓｉｏｎａｎｄＰａｔｔｅｒｎＲｅｃｏｇｎｉｔｉｏｎ（ＣＶＰＲ）．Ｈｏｎｏｌｕｌｕ，Ｈａｗａｉｉ：ＩＥＥＥ，２０１７：５９６－６０５．［７］ＳＵＢｏｌａｎ，ＬＵＳｈｉｊｉａｎ，ＴＡＮＣＬ．Ｂｌｕｒｒｅｄｉｍａｇｅｒｅｇｉｏｎｄｅｔｅｃｔｉｏｎａｎｄｃｌａｓｓｉｆｉｃａｔｉｏｎ［Ｃ］／／Ｐｒｏｃｅｅｄｉｎｇｓｏｆｔｈｅ１９ｔｈＡＣＭｉｎｔｅｒｎａｔｉｏｎａｌｃｏｎｆｅｒｅｎｃｅｏｎＭｕｌｔｉｍｅｄｉａ．ＳｃｏｔｔｓｄａｌｅＡｒｉｚｏｎａ，ＵＳＡ：ＡＣＭ，２０１１：１３９７－１４００．［８］ＳＨＩＪｉａｎｐｉｎｇ，ＸＵＬｉ，ＪＩＡＪｉａｙａ．Ｄｉｓｃｒｉｍｉｎａｔｉｖｅｂｌｕｒｄｅｔｅｃｔｉｏｎｆｅａｔｕｒｅｓ［Ｃ］／／ＰｒｏｃｅｅｄｉｎｇｓｏｆｔｈｅＩＥＥＥＣｏｎｆｅｒｅｎｃｅｏｎＣｏｍｐｕｔｅｒＶｉｓｉｏｎａｎｄＰａｔｔｅｒｎＲｅｃｏｇｎｉｔｉｏｎ．Ｃｏｌｕｍｂｕｓ，ＵＳＡ：ＩＥＥＥ，２０１４：２９６５－２９７２．［９］ＴＡＮＧＣｈａｎｇ，ＷＵＪｉｎ，ＨＯＵＹｏｎｇｈｏｎｇ，ｅｔａｌ．Ａｓｐｅｃｔｒａｌａｎｄｓｐａｔｉａｌａｐｐｒｏａｃｈｏｆｃｏａｒｓｅ－ｔｏ－ｆｉｎｅｂｌｕｒｒｅｄｉｍａｇｅｒｅｇｉｏｎｄｅｔｅｃｔｉｏｎ［Ｊ］．ＩＥＥＥＳｉｇｎａｌＰｒｏｃｅｓｓｉｎｇＬｅｔｔｅｒｓ，２０１６，２３（１１）：１６５２－１６５６．［１０］王雪玮．基于特征学习的模糊图像质量评价与检测分割研究［Ｄ］．合肥：中国科学技术大学，２０２０．［１１］ＣＨＥＮＱｉｆｅｎｇ，ＬＩＤｉｎｇｚｅｙｕ，ＴＡＮＧＣＫ，ｅｔａｌ．ＫＮＮＭａｔｔｉｎｇ［Ｊ］．ＩＥＥＥＴｒａｎｓａｃｔｉｏｎｓｏｎＰａｔｔｅｒｎＡｎａｌｙｓｉｓ＆ＭａｃｈｉｎｅＩｎｔｅｌｌｉｇｅｎｃｅ，２０１３，３５（９）：２１７５－２１８８．［１２］ＯＪＡＬＡＴ，ＰＩＥＴＩＫＡＩＮＥＮＭ，ＭＡＥＮＰＡＡＴ．Ｍｕｌｔｉｒｅｓｏｌｕｔｉｏｎｇｒａｙ－ｓｃａｌｅａｎｄｒｏｔａｔｉｏｎｉｎｖａｒｉａｎｔｔｅｘｔｕｒｅｃｌａｓｓｉｆｉｃａｔｉｏｎｗｉｔｈｌｏｃａｌｂｉｎａｒｙｐａｔｔｅｒｎｓ［Ｊ］．ＩＥＥＥＴｒａｎｓａｃｔｉｏｎｓｏｎＰａｔｔｅｒｎＡｎａｌｙｓｉｓａｎｄＭａｃｈｉｎｅＩｎｔｅｌｌｉｇｅｎｃｅ，２００２，２４（７）：９７１－９８７．［１３］ＡＣＨＡＮＴＡＲ，ＳＨＡＪＩＡ，ＳＭＩＴＨＫ，ｅｔａｌ．ＳＬＩＣｓｕｐｅｒｐｉｘｅｌｓｃｏｍｐａｒｅｄｔｏｓｔａｔｅ－ｏｆ－ｔｈｅ－ａｒｔｓｕｐｅｒｐｉｘｅｌｍｅｔｈｏｄｓ［Ｊ］．ＩＥＥＥＴｒａｎｓａｃｔｉｏｎｓｏｎＰａｔｔｅｒｎＡｎａｌｙｓｉｓａｎｄＭａｃｈｉｎｅＩｎｔｅｌｌｉｇｅｎｃｅ，２０１２，３４（１１）：２２７４－２２８２．［１４］ＷＡＮＧＪｉｎｇｄｏｎｇ，ＪＩＡＮＧＨｕａｉｚｕ，ＹＵＡＮＺｅｊｉａｎ，ｅｔａｌ．Ｓａｌｉｅｎｔｏｂｊｅｃｔｄｅｔｅｃｔｉｏｎ：Ａｄｉｓｃｒｉｍｉｎａｔｉｖｅｒｅｇｉｏｎａｌｆｅａｔｕｒｅｉｎｔｅｇｒａｔｉｏｎａｐｐｒｏａｃｈ［Ｊ］．ＩｎｔｅｒｎａｔｉｏｎａｌＪｏｕｒｎａｌｏｆＣｏｍｐｕｔｅｒＶｉｓｉｏｎ，２０１７，１２３：２５１－２６８．［１５］ＺＨＡＯＭｉｎｇｈｕａ，ＬＩＤａｎ，ＳＨＩＺｈｅｎｇｈａｏ，ｅｔａｌ．ＢｌｕｒｆｅａｔｕｒｅｅｘｔｒａｃｔｉｏｎｐｌｕｓａｕｔｏｍａｔｉｃＫＮＮｍａｔｔｉｎｇ：Ａｎｏｖｅｌｔｗｏｓｔａｇｅｂｌｕｒｒｅｇｉｏｎｄｅｔｅｃｔｉｏｎｍｅｔｈｏｄｆｏｒｌｏｃａｌｍｏｔｉｏｎｂｌｕｒｒｅｄｉｍａｇｅｓ［Ｊ］．ＩＥＥＥＡｃｃｅｓｓ，２０１９，７：１８１１４２－１８１１５１．［１６］ＯＴＳＵＮ．Ａｔｈｒｅｓｈｏｌｄｓｅｌｅｃｔｉｏｎｍｅｔｈｏｄｆｒｏｍｇｒａｙ－ｌｅｖｅｌｈｉｓｔｏｇｒａｍｓ［Ｊ］．ＩＥＥＥＴｒａｎｓａｃｔｉｏｎｓｏｎＳｙｓｔｅｍｓ，Ｍａｎ，ａｎｄＣｙｂｅｒｎｅｔｉｃｓ，１９７９，９（１）：６２－６６．［１７］ＧＡＳＴＡＬＥＳＬ，ＯＬＩＶＥＩＲＡＭＭ．Ｄｏｍａｉｎｔｒａｎｓｆｏｒｍｆｏｒｅｄｇｅ－ａｗａｒｅｉｍａｇｅａｎｄｖｉｄｅｏｐｒｏｃｅｓｓｉｎｇ［Ｊ］．Ｅｕｒｏｇｒａｐｈｉｃｓ，２０１０，２９（２）：７５３－７６２．（上接第１５０页）［４］ＫＡＺＥＭＩＦＭ，ＳＡＭＡＤＩＳ，ＰＯＯＲＲＥＺＡＨＲ，ｅｔａｌ．ＶｅｈｉｃｌｅｒｅｃｏｇｎｉｔｉｏｎｕｓｉｎｇｃｕｒｖｅｌｅｔｔｒａｎｓｆｏｒｍａｎｄＳＶＭ［Ｃ］／／Ｐｒｏｃ．ｏｆｔｈｅ４ｔｈＩｎｔｅｒｎａｔｉｏｎａｌＣｏｎｆｅｒｅｎｃｅｏｎＩｎｆｏｒｍａｔｉｏｎＴｅｃｈｎｏｌｏｇｙ．ＬａｓＶｅｇａｓ，ＮＶ，ＵＳＡ：ＩＥＥＥ，２００７：５１６－５２１．［５］ＦＲＥＵＮＤＹ，ＳＣＨＡＰＩＲＥＲＥ．Ａｄｅｃｉｓｉｏｎ－ｔｈｅｏｒｅｔｉｃｇｅｎｅｒａｌｉｚａｔｉｏｎｏｆｏｎ－ｌｉｎｅｌｅａｒｎｉｎｇａｎｄａｎａｐｐｌｉｃａｔｉｏｎｔｏｂｏｏｓｔｉｎｇ［Ｊ］．ＪｏｕｒｎａｌｏｆＣｏｍｐｕｔｅｒａｎｄＳｙｓｔｅｍＳｃｉｅｎｃｅｓ，１９９７，５５：１１９－１３９．［６］ＬＩＵＷｅｉ，ＥＲＨＡＮＤ，ＳＺＥＧＥＤＹＣ，ｅｔａｌ．ＳＳＤ：Ｓｉｎｇｌｅｓｈｏｔｍｕｌｔｉｂｏｘｄｅｔｅｃｔｏｒ［Ｃ］／／ＥｕｒｏｐｅｏｎＣｏｎｆｅｒｅｎｃｅｏｎＣｏｍｐｕｔｅｒＶｉｓｉｏｎ（ＥＣＣＶ）．Ｓｗｉｔｚｅｒｌａｎｄ：Ｓｐｒｉｎｇｅｒ，２０１６：２１－３７．［７］ＲＥＤＭＯＮＪ，ＤＩＶＶＡＬＡＳ，ＧＩＲＳＨＩＣＫＲ，ｅｔａｌ．Ｙｏｕｏｎｌｙｌｏｏｋｏｎｃｅ：Ｕｎｉｆｉｅｄ，ｒｅａｌ－ｔｉｍｅｏｂｊｅｃｔｄｅｔｅｃｔｉｏｎ［Ｃ］／／ＩＥＥＥＣｏｎｆｅｒｅｎｃｅｏｎＣｏｍｐｕｔｅｒＶｉｓｉｏｎａｎｄＰａｔｔｅｒｎＲｅｃｏｇｎｉｔｉｏｎ．ＬａｓＶｅｇａｓ，ＮＶ，ＵＳＡ：ＩＥＥＥ，２０１６：７７９－７８８．［８］ＲＥＤＭＯＮＪ，ＦＡＲＨＡＤＩＡ．ＹＯＬＯｖ３：Ａｎｉｎｃｒｅｍｅｎｔａｌｉｍｐｒｏｖｅｍｅｎｔ［Ｊ］．ａｒＸｉｖｐｒｅｐｒｉｎｔａｒＸｉｖ：１８０４．０２７６７，２０１８．［９］ＬＩＮＴＹ，ＧＯＹＡＬＰ，ＧＩＲＳＨＩＣＫＲ，ｅｔａｌ．Ｆｏｃａｌｌｏｓｓｆｏｒｄｅｎｓｅｏｂｊｅｃｔｄｅｔｅｃｔｉｏｎ［Ｃ］／／ＩＥＥＥＩｎｔｅｒｎａｔｉｏｎａｌＣｏｎｆｅｒｅｎｃｅｏｎＣｏｍｐｕｔｅｒＶｉｓｉｏｎ（ＩＣＣＶ）．Ｖｅｎｉｃｅ，Ｉｔａｌｙ：ＩＥＥＥ，２０１７：２９８０－２９８８．［１０］ＧＩＲＳＨＩＣＫＲ．ＦａｓｔＲ－ＣＮＮ［Ｃ］／／ＩＥＥＥＩｎｔｅｒｎａｔｉｏｎａｌＣｏｎｆｅｒｅｎｃｅｏｎＣｏｍｐｕｔｅｒＶｉｓｉｏｎ（ＩＣＣＶ）．Ｓａｎｔｉａｇｏ，Ｃｈｉｌｅ：ＩＥＥＥ，２０１５：１４４０－１４４８．［１１］ＲＥＮＳｈａｏｑｉｎｇ，ＨＥＫａｉｍｉｎｇ，ＧＩＲＳＨＩＣＫＲ，ｅｔａｌ．ＦａｓｔｅｒＲ－ＣＮＮ：Ｔｏｗａｒｄｓｒｅａｌｔｉｍｅｏｂｊｅｃｔｄｅｔｅｃｔｉｏｎｗｉｔｈｒｅｇｉｏｎｐｒｏｐｏｓａｌｎｅｔｗｏｒｋｓ［Ｊ］．ＩＥＥＥＴｒａｎｓａｃｔｉｏｎｓｏｎＰａｔｔｅｒｎＡｎａｌｙｓｉｓａｎｄＭａｃｈｉｎｅＩｎｔｅｌｌｉｇｅｎｃｅ，２０１６，３９（６）：１１３７－１１４９．［１２］ＢＯＣＨＫＯＶＳＫＩＹＡ，ＷＡＮＧＣＹ，ＬＩＡＯＨＹＭ．ＹＯＬＯｖ４：Ｏｐｔｉｍａｌｓｐｅｅｄａｎｄａｃｃｕｒａｃｙｏｆｏｂｊｅｃｔｄｅｔｅｃｔｉｏｎ［Ｊ］．ａｒＸｉｖｐｒｅｐｒｉｎｔａｒＸｉｖ：２００４．１０９３４，２０２０．［１３］ＡＺＨＡＲＭＩＨ，ＺＡＭＡＮＦＨＫ，ＴＡＨＩＲＮＭ，ｅｔａｌ．ＰｅｏｐｌｅｔｒａｃｋｉｎｇｓｙｓｔｅｍｕｓｉｎｇＤｅｅｐＳＯＲＴ［Ｃ］／／１０ｔｈＩＥＥＥＩｎｔｅｒｎａｔｉｏｎａｌＣｏｎｆｅｒｅｎｃｅｏｎＣｏｎｔｒｏｌＳｙｓｔｅｍ，ＣｏｍｐｕｔｉｎｇａｎｄＥｎｇｉｎｅｅｒｉｎｇ（ＩＣＣＳＣＥ）．ＩＥＥＥ，２０２０：１３７－１４１．７５１第４期李浩伟，等：基于ＬＢＰ特征与图像显著性的散焦模糊区域检测。

Word Translation警告标志Warning sign禁令标志Prohibition sign指路标志Guide sign旅游标志Tourist sign国道National road省道Provincial road县道County road服务区Service area轮渡Ferry车道封闭Lane closed车辆慢行Slow down双向交通Two-way traffic交通规则traffic regulation里程碑milestone红绿灯traffic light断头路，死胡同dead end速度限制speed limit路旁，设路旁，迂回bypass停车场car parking车载的car-borne车道，马路carriageway十字路crossroad十字路口intersection自行车道cycle track设计视距design sight distance起讫点调查OD--------origin and destination survey通行权，路权right-of–way环境污染environmental pollution功能分级functional classes分离式立体交叉口grade-separate d junctions立体交叉口Grade-separated junc tions环形交叉口roundabout intersecti ons实时自适应控制系统SCATS：Sydney Co-ordinated Adaptive t raffic system绿信比-信号周期-绿时差优化技术S COOT：split cycle-offset optimization te chnique城市公共交通系统urban public transport system地下铁道，地铁subway公共交通优先public transport priorityUnit1Transport n.&v. 运输Urban transportation planning 城市运输规划Transport demand 运输需求Transport modes 运输模式Mobility n. 机动性Energe-supply 能源供应Operating costs 运营成本Haul /tow 拖Both direct and indirect effects 直接和间接影响Initial and recurrent cost 初始和周期成本Vehicle operating costs 车辆运营费用Vehicle operating-cost saving 节省车辆运营费用Saving in travel time 节省运行时间Accident costs 事故成本Accident reduction 减少事故Time saving 省时Freeway 高速公路Expressway 快速路Highway 公路Port 港口Deep-water port 深水港Airport /terminal 机场/候机厅Airline cost 航线成本Rail 铁路Signalized street 由信号控制的街道Earth/gravel/bitumen 土/砂砾/沥青Aggregate 骨料Cement 水泥Asphalt / bitumenBlock 木块，石块Brick 砖块Mortar 灰浆，砂浆PlasticsWoodAdmixtureIn cementIn concreteIn mortarAsphalt –aggregate mixture （沥青骨料混合物）Asphalt (asphaltic) binder 沥青结合料Asphalt concrete(pavement) /bitulithic/bituminous concrete pavement沥青混凝土（路面）Asphalt covering/carpeting 沥青面层Asphalt/ bituminous macadam 沥青碎石Base 基层bituminous Mat 沥青垫层cement and sand cushion 水泥砂浆垫层Unit2V orad----Vehicle On-board Radar 车载雷达系统CHAUFFEUR （法）司机Longitude collision avoidance 纵向防撞Lateral collision avoidance 横向防撞Vision enhancement 防撞视野强化Safety readiness 危险预警Pre-crash restraint deployment 撞前避伤Automated vehicle operation 自动机车操作Warning systems 预警系统Forward collision warning 前防撞预警Truck lane departure warning systems Short-range warning of parking hazardsAdaptive cruise control 自适应式定速巡航系统Collision avoidance braking 防撞制动Full automation systems 全自动系统Urban transit systems 城市运输系统On-board sensors 车载感应器infrared LED taillight 红外液晶显示尾灯Fluorescent paint striping 银光涂料车道标线Smartway 智能道路Intersection collision warning 交叉口防撞预警Information processing 信息处理Control 控制Electronics 电子学Communications 通信Controlling technology 控制技术Four-way traffic signal 四路交通信号Traffic accidents and congestion 交通事故和交通拥挤Intelligent transportation society of America（ITS America) 美国智能运输系统协会Automatically collecting tolls 自动收费系统Rerouting traffic flow 交通流改道Weigh-in-motion systems 动态称重系统Automated tracking 自动跟踪Destination 目的地Navigation systems 导航系统Route guidance 指路Sight distanceDiesel 柴油机police car 警车Ambulance 急救车Motorcycle 摩托车，机车Motorway 汽车专用道number plate survey 车牌号调查one-way street 单向交通stopping sight distance 停车视距three-way junctions 三路交叉口Unit4public transport priority 公交优先urban transport strategy 城市交通发展战略Bus 公共汽车Tram 有轨电车bus route corridors 公交路线走廊traffic signal control 交通信号控制bus stop 公交车站with-flow 顺流Contraflow 逆车流方向，反向车流no-entry 禁止进入banned turn 禁止掉头traffic metering 咪表car-borne 车载的Design speed 设计车速Average running speed 平均行驶车速Operating speed 运行车速Average overall travel speed 平均行程车速Control of access 进口控制Full control 全控制Partial control 部分控制Curvature 曲率Superelevation 超高Gradient 坡度Major rural highways 主要乡村道路Vehicle and Pedestrians 车辆与行人Accident hazard 事故灾害Two-way left-turn lane 双向左转车道Initial allowance 预留Unit5urban transit systems 城市运输系统motor vehicle 机动车railroad engineering 铁路工程right- of- way 路权urban transit =mass transitMRT= Mass rapid transit Paratransit 辅助客运系统Fixed/established routes and fixed schedules 固定路线和行程时间表Taxisvanpool//carpool// 拼车，共用车辆club busesvanMinibusTrolley bus 电车Articulated bus 铰接车charter service 租赁服务High-occupancy-vehicle 高占有率的车辆Rolling stock 全部车辆Passenger transportation 客运Guideway 导沟，导轨Internal combustion engine 内燃发动机A diesel engine 柴油发动机Local transit 当地公交（客运运输）系统Express service 快速交通服务Peak service 高峰小时服务Short-haul transit 短途拖车运输Reading materialRail transit 轨道运输Exclusive facilities 专用设施Turning radii 转弯半径Sharp curve 急弯Commuter railroad 市郊列车Unit8Sight distance 视距Passing sight distance 超车视距Design sight distance 设计视距Stopping sight distance 停车视距Street 街道Two-lane rural highways 双车道乡村公路Two-lane two-way highways 双向双车道公路Arterial 主干道The opposing traffic lane 对向车道Overtaken vehicle 超载车辆Brakes reaction distance 反应距离Braking distance 制动距离Cross the centerline of four-lane undivided roadways or crossing the median of four lane divided roadways 穿越无分割的四车道的道路中线或是跨越分隔四车道的中央分割带Reading materialHorizontal and vertical alignment 平面与纵断面线形Economically feasible 经济可行Design speed 设计车速Curvature 曲率Laws of mechanics 力学原理（定律）Centrifugal force 离心力The vehicle weight component 车辆重力分力Roadway superelevation 道路超高Side friction 横向摩阻力Side friction factor 横向摩阻系数At constant speed =at the same speed Terrain 地形Level Terrain 平原区Rolling Terrain 丘陵区Mountainous Terrain 山岭区Steep slopes= abruptRise above and fall below 起伏Grades 纵坡Geometric features 几何特征Unit 15Traffic planning 交通规划Transportation planning 运输规划Traffic mode 交通模式Transportation system 运输系统Traffic flow simulation 交通流模拟Planning process 规划程序Data collection 数据采集Forecast 预测Socioeconomic impact estimation 社会经济影响评价Energy/environmental impact estimation 能源/环境影响评估Policy planning 政策规划Systems planning 系统规划Corridor planning 高速通道走廊规划Preliminary engineering 初期工程Engineering design 工程设计Human resources 人力资源Crude oil 原油Slurry 泥浆Gas 天然气Petroleum product 石油产品Road 道路Railway 铁路Transit line 运输线路Bus line 公交线路Waterway 水路Pipeline 管道Transportation congestion 交通运输拥挤Transport mode 交通运输模式Transportation survey 交通运输调查Trip generation 出行产生Trip distribution 出行分布Traffic assignment 交通分配Modal split 交通方式划分Unit 16 Community viewpoint 公众观点Urban transportation planning 城市运输规划Transportation facilities and operations 运输设施和运营The ratio of transit fares to personal income 交通费和个人收入之比Tabulation of O-D matrix O-D矩阵表Trip origins 起点Trip attractions 出行吸引量Fratar method 福雷特法Intervening opportunities model 插入机会模型Gravity model 重力模型Target planning year 目标规划年（远景规划年）Subroutine 子程序All-or-nothing assignment 全有全无分配法Capacity restrained assignment 交通容量限制分配法Multiple proportional assignment 多路径概率分配法Trip ends 出行端点The route of least cost 最少费用路径Traffic zones 交通区Unit 17Highway capacity 公路通行能力Level of services 服务水平At or near capacity 达到或接近通行能力Ideal condition 理想条件Uninterrupted flow 非间断流Intersection approaches 接近交叉口处Traffic stream 交通流Curb parking 路边停车Non-central business district area 非商务中心区Signalized intersection 有信号灯控制的交叉口Service flow rate 服务流率Level of service 服务水平Crawl speeds 爬坡车速Traffic stream 交通流Median 中央分割带Signalized intersections 信号交叉口Vehicle per hour（vph）辆/小时Left, through , right 左转、直行、右转Pedestrian crossing flows 过街行人流量Signal phasing 信号相位Timing 信号配时Type of control 控制类型Saturation flow 饱和流量Saturation flow rate 饱和流率Effective green time 有效绿灯时间Vehicle per hour of Effective green time （vphg）辆/有效绿灯小时Start-up lost time 启动损失时间Cycle length 周期长road marking 道路标线traffic sign regulations 交通标志规则。

> GPS+GSM+SMS/GPRS+OBD II <使用说明书User ManualV1.0O B D I I车载定位终端感谢您选用购买本机器，请您在使用之前认真阅读本说明书，以便得到正确的安装方法及操作指南，以下描述中终端等同于本机器。

产品外观及配色如有改动，请以实物为准，恕不另行通知。

本车用定位跟踪产品借助GPS卫星定位、OBD自动诊断系统、GPRS通信、Internet，通过强大的WEB服务平台可以实现对车辆进行实时远程定位监控和远程汽车诊断。

帮助客户实现透明管理、降低成本、保障安全、提高效率的目标。

目前已广泛应用于商业运输、物流配送、企业车队、汽车租赁、智能交通、工程机械、船舶航运、应急指挥、抢险施救、军警安监、智慧城市。

本设备分3种不同功能型号以适应不同需求：A：GPS+OBDB：GPS+电子狗C：GPS目 录一、功能特点 (7)二、构件名称 (7)三、使用环境 (9)四、基本参数 (9)五、安装说明 (9)5.1安装前准备事项 (9)5.2 SIM卡的安装 (10)六、配置终端 (12)七、使用终端 (12)7.1开机 (12)7.2指示灯 (12)7.3查看位置 (14)7.3.1短信查询 (14)7.3.2终端服务平台查询 (14)7.4碰撞/跌落报警 (14)八、登录终端定位服务平台 (14)8.1浏览器平台 (14)8.2 智能手机客户端 (15)九、故障排除 (15)9.1无法连接服务平台 (15)9.2后台显示离线状态 (15)9.3长时间不定位 (16)9.4定位漂移严重 (16)9.5指令接收异常 (16)十、设备保修细则 (17)10.1特别声明 (17)10.2保修期 (17)10.3售后服务 (17)十一、操作指令 (18)User Manual (19)Ⅰ. Product Features (20)Ⅱ. Components & Accessories (21)2.1 Components (21)2.2 Accessories（reference pictures）22 Ⅲ. Environment for use (22)Ⅳ. Basic Specifications (23)Ⅴ. Installation (23)5.1 Before Installation (23)5.2 Install the SIM card (24)Ⅵ. Set up the terminal (26)Ⅶ. Use the terminal (26)7.1 Power on (26)7.2 LED indicators (27)7.3 Inquiry position (28)7.4 Collision / falling Alarm (28)Ⅷ. Login the position server (29)Ⅸ. Trouble shooting (29)9.1 Cannot connect to the position server (29)9.2 Show offline status on the positionserver (30)9.3 Cannot position for a long time (31)9.4 Position drift (31)9.5 Instructions receiving abnormally.31 Ⅹ. Warranty rules (32)10.1 Special statement (32)10.2 Warranty period (32)10.3 After sales (32)Ⅻ. Operation Commands (33)Customer’s Information (34)保修卡资料 (34)Maintenance records / 维修记录 (35)一、功能特点■GSM四频系统，全球通用■安装极方便，连上OBD接口即可工作■GPS连续定位，OBD实时数据，GPRS定时上报■车辆定位追踪，浏览器、手机、短信远程查询■OBD转速检测，车辆参数，精度超ACC（A款）■电子狗语音播报（B款）■车辆发生碰撞、跌落时通过短信、平台报警二、构件名称-机身正面--机身背面-配件名称（以下图片仅供参考，以实物为准）OBD II延长线此配件为设备选配，当OBD接口位置不佳导致GPS信号接收不好的时候，可加配此延长线。

空车走行率名词解释英文Title: An Explanation of the Term "Empty Car Travel Rate"The term "empty car travel rate" is a metric commonly used in the transportation industry, particularly in rail and freight transport. It refers to the percentage of total trips made by a vehicle that are without passengers or cargo. In other words, it measures the efficiency of a vehicle's use by assessing how often it is traveling without carrying its intended load.This metric is important for several reasons. Firstly, it provides a quantifiable measure of the efficiency of a transportation system. If a high proportion of trips are made with empty vehicles, it suggests that there may be inefficiencies in the system, such as imbalanced demand and supply, routing issues, or insufficient utilization of available resources.Secondly, the empty car travel rate can help transportation providers identify areas for improvement. By analyzing the factors that contribute to high empty travel rates, companies can make more informed decisions about routing, scheduling, and capacity management. This can lead to cost savings, increased efficiency, and better customer satisfaction.Thirdly, the empty car travel rate is also relevant for environmental considerations. Reducing the number of empty trips can help to decrease fuel consumption and greenhouse gas emissions, which are significant contributors to climate change. By optimizing vehicle utilization, transportation companies can make a positive contribution to sustainable development.To calculate the empty car travel rate, the total number of trips made by a vehicle is divided by the number of trips that are carried out with a full load. The resulting percentage provides a measure of the efficiency of vehicle utilization. While the ideal empty travel rate would be as low as possible, it is important to note thatsome empty trips may be necessary to maintain theefficiency of the transportation system, such as repositioning vehicles between routes or to service maintenance facilities.There are several factors that can contribute to high empty car travel rates. Imbalanced demand and supply, where the availability of vehicles does not match the demand for transportation services, can lead to a higher proportion of empty trips. Routing issues, such as inefficient routing algorithms or a lack of real-time data to optimize routes, can also contribute to empty travel. Additionally, insufficient utilization of available resources, such as underutilized vehicles or unoccupied seats on public transport, can lead to increased empty travel rates.To address these issues and improve vehicle utilization, transportation providers can implement a number of strategies. One approach is to use advanced analytics and data-driven decision-making tools to optimize routing and scheduling. By analyzing historical data and real-time information, companies can better predict demand patternsand adjust their operations to match supply and demand more closely.Another strategy is to enhance the use of technology to improve vehicle tracking and management. GPS tracking systems, for example, can provide real-time information on vehicle locations and loads, enabling companies to makemore informed decisions about routing and repositioning. Similarly, smart scheduling systems can help to optimizethe allocation of vehicles and reduce the number of empty trips.Moreover, transportation providers can collaborate with other companies or organizations to share resources and improve the efficiency of the overall transportation system. For instance, ride-sharing or freight-sharing platforms can match supply and demand more effectively, reducing the need for empty trips. Similarly, intermodal transportation systems that combine different modes of transport, such as rail, road, and water, can improve the efficiency of the overall network and reduce empty travel.In conclusion, the empty car travel rate is a critical metric for assessing the efficiency of transportation systems. By understanding the factors that contribute to high empty travel rates and implementing strategies to address them, transportation providers can improve their operations, reduce costs, and make a positive contribution to sustainable development. As the demand for transportation services continues to grow, it will be increasingly important for companies to prioritize efficiency and sustainability in their operations.。

汽车移动速度测量方法的探究摘要：本文简要介绍了几种在现实生活中常用的测量汽车移动速度的方法，着重分析了激光雷达测速原理，推导了连续激光脉冲数字、多普勒频移测速的方法，给出车载激光雷达基本原理图，为车载激光雷达系统测距测速提供了基本方法。

关键字：激光雷达，测距，测速Vehicle Speed Measurement Methods of InquiryAuthor:DongHaoTutor:BaiXiaoLei Abstract：This paper briefly introduces several kinds of commonly used in real life the measurement method of vehicle speed, speed measuring principle of laser radar is analyzed emphatically, laser pulse number is derived, the doppler frequency shift speed measuring method, gives the principle diagram of the vehicular laser radar basic, for vehicular laser ranging radar system provides a basic method of speed.Key words：laser radar, range, speed目录绪论 (1)一、交通中常用的测量移动速度的方法 (2)1. 雷达测速 (2)2. 激光测速 (2)3. 视频测速 (2)4. IC卡测速 (2)二、目标距离测量原理 (3)1. 测距原理 (3)2. 测距方法的选择 (3)三、目标相对速度的测量原理 (5)1. 相对速度测量原理 (5)2. 相对速度的测量方法 (7)结语 (8)参考文献 (9)绪论随着我国高速公路网的不断完善，高速公路给人们带来了交通便利和能源的节省，同时车辆在高速公路上超速行驶也会容易引发交通事故。