[{"data":1,"prerenderedAt":228},["ShallowReactive",2],{"method-dtam2011":3},{"method":4,"reference":47,"equipment":67,"figures":82,"results":83},{"id":5,"label":6,"shortName":7,"title":8,"year":9,"era":10,"cluster":11,"scope":12,"keyIdeaZh":13,"keyIdeaEn":14,"fulltextStatus":15,"publicationStatus":16,"recommendation":17,"constructionRelevance":18,"validationEnvironment":19,"strengths":21,"limitations":23,"sensors":30,"platform":32,"estimator":34,"association":35,"timeModel":36,"deskew":37,"loopClosure":38,"globalOptimization":39,"mapRepresentation":40,"prior":41,"outputGeometry":42,"compute":43,"codeUrl":44,"codeLicense":45,"relatedVersions":46},"dtam2011","Newcombe et al., 2011a","DTAM","DTAM: Dense tracking and mapping in real-time",2011,"classic","C08","odometry_with_local_mapping","DTAM 不擷取特徵點，而是以每個像素的光度資料在關鍵影格上估計稠密深度圖，並以空間正則化能量函數求解，形成大量頂點的表面拼貼。相機位姿則以整張影像對稠密模型進行直接對齊（direct alignment）追蹤。此方法依賴 GPU 平行運算，並假設靜態場景。","DTAM reconstructs dense keyframe depth maps from every pixel with a regularised photometric energy and tracks the monocular camera by whole-image alignment to the dense model on a GPU.","full_text_reviewed","peer_reviewed_published","background","論文未報告營建或建築量測測試。對本文作為單眼稠密直接法的技術背景，說明稠密表面並不等同已驗證的工程尺度幾何（單眼尺度不可觀）。論文唯一的量化評估是與 PTAM 的線速度曲線比較，未提供地面真值軌跡或幾何精度數值（Sec. 3.1、Fig. 9），因此不能支撐任何工程精度主張；作者並指出系統假設亮度恆定，無法處理真實環境的全域光照變化（Sec. 3.2），因此不適合光照變化大的工地（推論）。",[20],"controlled_experiment",[22],"Dense model reported to give better tracking under rapid motion than a feature-based method (abstract)",[24,25,26,27,28,29],"Requires GPU hardware (abstract)","Assumes a static scene (abstract)","ORB-SLAM authors note direct dense methods such as DTAM only incrementally expand the map rather than jointly optimising it (orbslam2015, Sec. IX-B)","Assumes brightness constancy; not robust to real-world global illumination changes (Sec. 3.2 Failure Modes and Future Work)","Initialisation still relies on a feature-based stereo method; a fully dense initialisation is future work (Sec. 2.4)","(observation) Quantitative evaluation is limited to a velocity-profile comparison with PTAM; no ground-truth trajectory or geometric accuracy is reported (Sec. 3.1, Fig. 9)",[31],"monocular camera",[33],"hand-held Point Grey Flea2 RGB camera in a desktop setting (same setting where PTAM succeeded)","Mapping: per-keyframe inverse depth map minimising a photometric cost volume (average L1 error over tens to hundreds of overlapping frames at S inverse-depth samples) plus an edge-weighted Huber regulariser; the energy is decoupled with an auxiliary variable, solved by primal-dual updates for the convex part and a point-wise exhaustive search over the cost volume whose feasible range shrinks each iteration, with one embedded Newton step for sub-sample accuracy (theta from 0.2 to 1e-4). Tracking: Lucas-Kanade style iterative least squares, first inter-frame rotation on coarse pyramid levels, then 6DOF forward-compositional alignment of the live image to a view synthesised from the dense model, coarse to fine.","direct photometric every-pixel association: cost volume built by projecting reference pixels into overlapping frames for each inverse-depth sample; tracking compares every pixel of the live image with the model-predicted image, rejecting pixels whose photometric error exceeds a threshold that decreases during coarse-to-fine iterations","discrete poses","not_applicable","none described; the method has no loop detection or map correction, and a relocaliser is mentioned only as disabled during the PTAM comparison","none; each keyframe's regularised inverse depth map is estimated on its own (no joint optimisation across keyframes is described), and new keyframes are added according to a threshold on the number of pixels in the previous predicted image without visible surface information","overlapping keyframes, each with an RGB reference image, pose, inverse depth map and an M x N x S photometric cost volume; the Fig. 3 example keyframe has nearly 300 x 10^3 estimated points versus about 1000 PTAM point features in the same frame","pre-calibrated fixed intrinsics with images pre-warped to remove radial distortion; bootstrapped by a standard point-feature stereo method until the first keyframe; static scene and brightness constancy assumed","textured dense inverse depth maps; a triangle mesh is computed from each keyframe depth map (oblique edges culled) and used for tracking, forming a surface patchwork with millions of vertices","real-time on commodity GPU (abstract); experiments on an NVIDIA GTX 480 (Sec. 3)",null,"not_verified",[],{"id":5,"kind":48,"shortName":7,"title":8,"authors":49,"year":9,"venue":53,"venueType":54,"publisher":55,"volumeIssuePages":56,"doi":57,"arxivId":44,"url":58,"firstPublicDate":59,"publicationStatus":16,"metadataStatus":60,"fulltextStatus":15,"era":10,"classicReason":61,"codeUrl":44,"cluster":11,"topics":62,"mdpi":63,"verification":64,"label":6,"fulltextRoute":65,"versionRead":66,"addedByCensus":63},"method",[50,51,52],"Richard A. Newcombe","Steven J. Lovegrove","Andrew J. Davison","2011 International Conference on Computer Vision (ICCV)","conference","IEEE","pp. 2320-2327","10.1109\u002Ficcv.2011.6126513","https:\u002F\u002Fwww.doc.ic.ac.uk\u002F~ajd\u002FPublications\u002Fnewcombe_etal_iccv2011.pdf","2011-11","metadata_verified","necessary technical node: dense, every-pixel direct monocular tracking and keyframe depth-map reconstruction, the dense counterpart discussed by ORB-SLAM and LSD-SLAM.",[11],false,"corrected","author copy","author-hosted PDF of the ICCV 2011 paper (8 pages, Imperial College author page); IEEE Xplore version of record not compared",[68,74,79],{"category":69,"model":70,"canonical":70,"role":71,"dataset":44,"specs":72,"locator":73},"camera","Point Grey Flea2","method input","30 Hz, 640x480, 24-bit RGB colour, pre-calibrated intrinsics","Sec. 3",{"category":75,"model":76,"canonical":76,"role":77,"dataset":44,"specs":78,"locator":73},"compute","NVIDIA GTX 480","compute for runtime","commodity GPU running mapping and tracking",{"category":75,"model":80,"canonical":80,"role":77,"dataset":44,"specs":81,"locator":73},"i7 quad-core CPU","host CPU of the GPU system",[],{"totalRows":84,"groupCount":84,"groups":85,"others":227},1,[86],{"slug":87,"group":88,"sourceId":89,"sourceLabel":90,"table":91,"selfRows":84,"metrics":92,"seqs":97,"entrants":101,"cells":162,"outcomes":211,"locators":222,"hardware":223,"wordings":224,"notes":225},"ghadimzadeh2025slamnde-table-2","ghadimzadeh2025slamnde:Table 2","ghadimzadeh2025slamnde","Ghadimzadeh Alamdari et al., 2025","Table 2",[93],{"label":94,"unit":95,"statistic":96,"alignment":95},"Result (run outcome)","none","not_reported",[98],{"dataset":99,"sequence":96,"environment":100},"Luleå SubT tunnel dataset (Koval et al. 2022)","underground tunnel",[102,106,109,111,113,116,117,120,123,126,129,131,134,136,138,140,143,145,148,150,152,155,157,160],{"name":103,"methodId":104,"linkable":105,"proposed":63,"self":63},"Mono-SLAM","monoslam2007",true,{"name":107,"methodId":108,"linkable":105,"proposed":63,"self":63},"PTAM","ptam2007",{"name":110,"methodId":44,"linkable":63,"proposed":63,"self":63},"S-PTAM",{"name":112,"methodId":44,"linkable":63,"proposed":63,"self":63},"OV2SLAM",{"name":114,"methodId":115,"linkable":105,"proposed":63,"self":63},"ORB-SLAM (footnote 1)","orbslam2015",{"name":7,"methodId":5,"linkable":105,"proposed":63,"self":105},{"name":118,"methodId":119,"linkable":105,"proposed":63,"self":63},"LSD-SLAM","lsdslam2014",{"name":121,"methodId":122,"linkable":105,"proposed":63,"self":63},"SVO","svo2017",{"name":124,"methodId":125,"linkable":105,"proposed":63,"self":63},"DSO","dso2018",{"name":127,"methodId":128,"linkable":105,"proposed":63,"self":63},"Kinetic Fusion","kinectfusion2011",{"name":130,"methodId":44,"linkable":63,"proposed":63,"self":63},"Dense visual SLAM",{"name":132,"methodId":133,"linkable":105,"proposed":63,"self":63},"Elastic Fusion SLAM","elasticfusion2015",{"name":135,"methodId":44,"linkable":63,"proposed":63,"self":63},"Realtime onboard VI estimation",{"name":137,"methodId":44,"linkable":63,"proposed":63,"self":63},"Multi-sensor fusion",{"name":139,"methodId":44,"linkable":63,"proposed":63,"self":63},"SOFT-SLAM",{"name":141,"methodId":142,"linkable":105,"proposed":63,"self":63},"MSCKF","mourikis2007msckf",{"name":144,"methodId":44,"linkable":63,"proposed":63,"self":63},"ROVIO",{"name":146,"methodId":147,"linkable":105,"proposed":63,"self":63},"OKVIS","okvis2015",{"name":149,"methodId":44,"linkable":63,"proposed":63,"self":63},"VIORB",{"name":151,"methodId":44,"linkable":63,"proposed":63,"self":63},"S-MSCKF",{"name":153,"methodId":154,"linkable":105,"proposed":63,"self":63},"VINS-Mono","vinsmono2018",{"name":156,"methodId":44,"linkable":63,"proposed":63,"self":63},"STCM-SLAM",{"name":158,"methodId":159,"linkable":105,"proposed":63,"self":63},"Kimera","kimera2020",{"name":161,"methodId":44,"linkable":63,"proposed":63,"self":63},"Yolo-SLAM",[163,166,167,169,171,173,175,177,179,181,183,185,187,189,191,193,195,197,199,201,203,205,207,209],[164,164,164,44,164,164,165,165,164],0,-1,[84,164,164,44,84,164,165,165,164],[168,164,164,44,168,164,165,165,164],2,[170,164,164,44,168,164,165,165,164],3,[172,164,164,44,170,164,165,165,164],4,[174,164,164,44,172,164,165,165,164],5,[176,164,164,44,174,164,165,165,164],6,[178,164,164,44,176,164,165,165,164],7,[180,164,164,44,168,164,165,165,164],8,[182,164,164,44,172,164,165,165,164],9,[184,164,164,44,164,164,165,165,164],10,[186,164,164,44,178,164,165,165,164],11,[188,164,164,44,172,164,165,165,164],12,[190,164,164,44,172,164,165,165,164],13,[192,164,164,44,172,164,165,165,164],14,[194,164,164,44,176,164,165,165,164],15,[196,164,164,44,168,164,165,165,164],16,[198,164,164,44,176,164,165,165,164],17,[200,164,164,44,178,164,165,165,164],18,[202,164,164,44,172,164,165,165,164],19,[204,164,164,44,180,164,165,165,164],20,[206,164,164,44,172,164,165,165,164],21,[208,164,164,44,164,164,165,165,164],22,[210,164,164,44,182,164,165,165,164],23,[212,213,214,215,216,217,218,219,220,221],"failed (feature detection and tracking)","failed (initialization for ground floor)","not_run (authors could not run the code)","success (footnote 1: authors could not run ORB-SLAM 3, so the original ORB-SLAM was used)","not_run (no publicly available repository)","failed (feature tracking)","failed (tracking)","not_run (inconsistent repository)","success","other: Result cell reads 'SLAM for dynamic environments'; no run outcome stated",[91],[],[],[226],"Run outcome ('Result' column) of each reviewed vision-based method on the Luleå tunnel test dataset; '+' marks methods not integrated with ROS; the '*' (incompatible with VLP-16) symbol is printed on almost every row",[],1790510658100]