％加载与地面真实数据的垫文件加载=负载(“caltech_cordova1_laneAndVehicleGroundTruth.mat”）;

％创建一个录像机对象以读取视频的帧。VideoName =加载.VideOname;filereader = Videoreader（VideoName）;地面事实数据被组织成一个时间表。gtdata = loaded.groundtruthdata;％显示地面真理数据的前几行。头（gtdata）

%读取视频的第一帧。帧= readFrame (fileReader);％提取第一帧中的所有车道点。lanepoints = gtdata.laneboundaries {1};%在第一帧中提取车辆边界框。车辆箱= gtdata.vehlicle {1};%叠加右车道点和车辆边界框。Frame = InsertMarker（帧，LanePoints {2}，'X'）;Frame = InsertObjectAnnotation（帧，'长方形'，车辆箱，'车辆'）;%在第一帧显示地面真实数据。图imshow（框架）

%设置monoSensorHelper处理视频。monocameraSensor =加载。Sensor;monosensorhelper = helpermonosensor（单声道血管传感器）;%创建新的时间表，使用相同的时间向量进行度量。测量=时间表（gtdata.time）;％设置用于保持车道边界和车辆数据的时间表列。numframes = floor（filereader.framerate * filereader.duration）;测量.LaneBoundaries = cell（numframes，2）;测量。vehicledetections = cell（numframes，1）;gtdata.lanesinvehiclecoord = cell（numframes，2）;%回放视频到t = 0，并创建一个帧索引保持当前%框架。filereader.currenttime = 0;FrameIndex = 0;%循环通过videfile直到没有新的帧。尽管hasFrame(fileReader) frameIndex = frameIndex+1;帧= readFrame (fileReader);％使用ProcessFrame方法计算检测。%可以用自定义车道检测方法替换此方法。检测= ProcessFrame（MonosensorHelper，Frame）;%存储估计的车道边界和车辆检测。测量。(detections.leftEgoBoundary LaneBoundaries {frameIndex} =.．.detections.rightEgoBoundary];测量。VehicleDetections {frameIndex} = detections.vehicleBoxes;为便于比较，将地面真实巷点转换为%车辆坐标系。gtPointsThisFrame = gtdata.LaneBoundaries {frameIndex};vehicle = cell(1, numel(gtPointsThisFrame));为ii = 1:numel(gtpointtsthisframe) vehiclePoints{ii} = imageToVehicle(monocamersensor, gtpointtsthisframe {ii});结束％储存在车辆坐标中表达的地面真相。gtdata。LanesInVehicleCoord {frameIndex} = vehiclePoints;结束

%把视频倒回t = 0。filereader.currenttime = 0;%读取视频的第一帧。帧= readFrame (fileReader);BirdseyeImage = Transportimage（Monosensorhelper.birdseyeconfig，框架）;在鸟瞰视图中提取第一帧的右车道点。firstFrameVehiclePoints = gtdata.LanesInVehicleCoord {1};pointsInBEV = vehicleToImage (monoSensorHelper。BirdsEyeConfig, firstFrameVehiclePoints {2});%在框架上叠加点。birdsEyeImage = insertMarker(birdsEyeImage, pointsInBEV，'X'，'尺寸'6）;%在鸟瞰图中显示变换的点。图imshow（birdseyeimage）

阈值= 0.25;%在米[nummatches，nummisses，numfalsepositive，作业] =.．.evaluateLaneBoundaries(测量。LaneBoundaries,.．.gtdata。LanesInVehicleCoord,.．.临界点）;DISP（['比赛数量：'，num2str（nummatches）]）;DISP（['错过的人数：'，num2str（nummisses）]）;DISP（[误报次数:num2str (numFalsePositives)]);

averageDistance = helperComputeLaneStatistics(测量。LaneBoundaries,.．.gtdata。LanesInVehicleCoord,.．.作业，@mean）;％绘制估计与地面真相之间的平均距离。图阻止(gtdata。时间,averageDistance) title(“估算值与地面真实值之间的平均距离”） 网格在ylabel (“米距离”)传说(的左边界，“右边界”）

％使用单纸片计算估计的点位置。[estVehiclePoints, estImagePoints] = helpergetrespondingpoints (monocamersensor，.．.测量.LaneBoundaries，.．.gtdata。LanesInVehicleCoord,.．.作业）;%在测量时间表中添加估计的车道。测量。EstimatedLanes = estImagePoints;测量。LanesInVehicleCoord = estVehiclePoints;%创建一个新的时间表，包含可视化所需的所有变量。名称= {'lanepoints'；'检测到Lanepoints'};类型= labeltype（{“行”；“行”});labeldefs = table（名称，类型，“VariableNames”，{“名字”，'类型'});(gtdata visualizeInFrame =时间表。时间,.．.gtdata.laneboundaries，.．.测量。最期待，.．.“VariableNames”、名称);%创建groundTruth对象。数据源= groundTruthDataSource (videoName);dataToVisualize = groundTruth(dataSource, labelDefs, visualizeInFrame);%在distanceData中保存上一节的所有结果。垫在%的临时文件夹。dataToLoad = [tempdir'distanceata.mat'];保存（数据罗拉，“monoSensorHelper”，“videoName”，'测量'，'gtdata'，“averageDistance”）;

功能stat = helpercomputelanstatistics (estModels, gtPoints, assignments, fcnHandle) numFrames = length(estModels);默认情况下，％使左右估计南部缺乏％ 数据。stat = NaN*ones(numFrames, 2);为Frameind = 1：NumFrames默认情况下，％使左右估计。stat(frameInd，:) = NaN*ones(2,1);为idx = 1:长度(estModels {frameInd})忽略假阳性赋值。如果作业{frameInd} (idx) = = 0继续；结束% estModelInFrame中的第k个边界匹配到第k个%元素索引的分配在gtPointsInFrame。这种模式下= estModels {frameInd} (idx);thisGT = gtPoints {frameInd}{作业{frameInd} (idx)};thisGTModel = driving.internal.piecewiseLinearBoundary (thisGT);如果意味着(thisGTModel.Points (:, 2)) > 0％ 左车道xpoints = thinggtmodel.points（：1）;Ydist =零（尺寸（xpoints））;为gtYPoints = thisgtmodel . computeborderymodel (xPoints(index));testYPoints = thisModel.computeBoundaryModel (xPoints(指数));yDist(指数)= abs (testYPoints-gtYPoints);结束stat（frameind，1）= fcnhandle（ydist）;别的％ 右侧车道xpoints = thinggtmodel.points（：1）;Ydist =零（尺寸（xpoints））;为gtYPoints = thisgtmodel . computeborderymodel (xPoints(index));testYPoints = thisModel.computeBoundaryModel (xPoints(指数));yDist(指数)= abs (testYPoints-gtYPoints);结束stat（frameind，2）= fcnhandle（ydist）;结束结束结束结束

功能[vehiclePoints, imagePoints] = helpergetrespondingpoints (monocamersensor, estModels, gtPoints, assignments) numFrames = length(estModels);imagePoints = cell(numFrames, 1);vehicle = cell(numFrames, 1);为Frameind = 1：NumFrames如果isempty(assignments{frameInd}) imagePointsInFrame = [];vehiclePointsInFrame = [];别的estmodelinframe = estmodels {frameind};gtpointsinframe = gtpoints {frameind};imagePointsinframe = cell（长度（estmodelinframe），1）;vertinpointsinframe = cell（长度（estmodelinframe），1）;为IDX = 1：长度（Estmodelinframe）忽略假阳性赋值。如果分配{frameind}（idx）== 0 imagepointsinframe {idx} = [nan nan];继续；结束% estModelInFrame中的第k个边界匹配到第k个%元素索引的分配在gtPointsInFrame。thismodel = estmodelinframe（Idx）;thisgt = gtpointsinframe {dispordments {frameind}（idx）};xpoints = thinggt（：，1）;ypoints = thismodel.computeboundarymodel（xpoints）;verthpointsinframe {idx} = [xpoints，ypoints];imagePointsinframe {idx} = vevicletoimage（monocamerasensor，[xpoints ypoints]）;结束结束车辆点{frameind} = verthingspointsinframe;imagePoints {frameind} = imagepointsinframe;％makeportpoints []而不是{}符合Toundtruth对象。如果isempty(imagePoints{frameInd}) imagePoints{frameInd} = [];结束如果isempty（verthpoints {frameind}）resplopitts {frameind} = [];结束结束结束