[{"data":1,"prerenderedAt":449},["ShallowReactive",2],{"method-curless1996volumetric":3},{"method":4,"reference":41,"equipment":60,"figures":77,"results":78},{"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":26,"platform":28,"estimator":31,"association":32,"timeModel":32,"deskew":32,"loopClosure":32,"globalOptimization":33,"mapRepresentation":34,"prior":35,"outputGeometry":36,"compute":37,"codeUrl":38,"codeLicense":39,"relatedVersions":40},"curless1996volumetric","Curless & Levoy, 1996","Volumetric range-image integration (TSDF origin; VRIP)","A volumetric method for building complex models from range images",1996,"classic","C12","map_representation_or_reconstruction","作者將每張已對齊的距離影像（range image）沿感測器視線轉成有號距離函數與權重，逐一加權累加到體素格網中，最後擷取零等值面成為三角網格；在特定假設下，此等值面在最小平方意義上最佳。體素另外標記為空、未觀測或近表面三種狀態，藉由空間雕刻（space carving）在空與未觀測區域的交界補面，產生無孔洞模型。","Fuses aligned range images into a cumulative weighted signed distance field along sensor lines of sight and extracts the zero isosurface, with space carving used to fill holes.","full_text_reviewed","peer_reviewed_published","main_body","原研究為小型物件掃描（龍、20 cm 高的快樂佛像、1.6 mm 鑽頭），未在營建場域驗證；作者僅在 Sec. 7 提到未來希望應用到地形與建築場景。作者自述的薄面增厚與尖角圓化限制，直接對應牆板、鋼構翼板與構件邊緣等營建幾何的尺寸風險；孔洞填補產生的面並非量測資料，用於營建成果時需與觀測面分開標示（推論）。",[20],"controlled_experiment",[22],"[\"Integrates many range images (up to 70) into seamless models of up to 2.6 million triangles (abstract).\", \"Incremental, order-independent updates and a least-squares-optimal isosurface under stated assumptions (orthographic sensor, independent errors along lines of sight) (Sec. 3, Appendix A sketch).\", \"RMS distance between original range points and the reconstructed dragon and Buddha surfaces is about 0.1 mm, roughly the scanner accuracy (Sec. 6).\", \"Behaves robustly on a 1.6 mm drill bit scanned from 12 orientations, where the zippering method fails catastrophically (Sec. 6, Fig. 9).\", \"Run-length encoding gives typical memory savings of 10:1 to 20:1 (Sec. 5.2.1).\"]",[24,25],"[\"Difficulty bridging sharp corners when no scan spans both adjoining surfaces (Sec. 7).\", \"Thin surfaces are problematic: distance ramps extending behind surfaces interfere across opposite sides, causing thickening of thin surfaces and rounding of sharp corners (Sec. 7).\", \"Hole-fill surfaces are not observed geometry","without extra backdrop carving they create extraneous tessellations, and their unseen to empty transition produces aliasing that needs separate filtering (Sec. 4, Fig. 11).\", \"Optical scanning limits: only external surfaces are seen, shiny, dark or bright surfaces cause errors, and the authors often paint objects flat gray to reduce these effects (the Buddha was painted matte gray, the drill bit white) (Sec. 7, Figs. 9 and 12).\", \"Range images must be pre-aligned and weights follow optical-triangulation uncertainty (Sec. 3).\"]",[27],"[\"Cyberware 3030 MS laser-stripe optical triangulation scanner (traditional triangulation or spacetime analysis)\"]",[29,30],"[\"laser-stripe optical triangulation scanner with the object translated through the laser plane (the Fig. 5 caption calls the Cyberware a translating sensor)","objects repositioned between scans by a motion control platform\"]","not_applicable (range images are assumed pre-aligned)","not_applicable","none","cumulative weighted signed distance function on a voxel grid (run-length encoded) with empty\u002Funseen\u002Fnear-surface voxel states","aligned range images","watertight triangle mesh extracted as the zero isosurface; hole filling by tessellating between empty and unseen regions","offline; 250 MHz MIPS R4400: observed-surface integration under 1 h (47 to 56 min) and space carving with hole filling 3 to 5 h (197 to 257 min); up to 12 million input vertices and grids up to 160 million voxels",null,"not_verified",[],{"id":5,"kind":42,"shortName":7,"title":8,"authors":43,"year":9,"venue":46,"venueType":47,"publisher":48,"volumeIssuePages":49,"doi":50,"arxivId":38,"url":51,"firstPublicDate":52,"publicationStatus":16,"metadataStatus":53,"fulltextStatus":15,"era":10,"classicReason":54,"codeUrl":38,"cluster":11,"topics":55,"mdpi":56,"verification":57,"label":6,"fulltextRoute":58,"versionRead":59,"addedByCensus":56},"method",[44,45],"Brian Curless","Marc Levoy","Proceedings of the 23rd Annual Conference on Computer Graphics and Interactive Techniques (SIGGRAPH '96)","conference","ACM","pp. 303-312","10.1145\u002F237170.237269","https:\u002F\u002Fgraphics.stanford.edu\u002Fpapers\u002Fvolrange\u002Fvolrange.pdf","1996-08","metadata_verified","principle reused: the cumulative weighted signed distance function fused per voxel is the basis of later TSDF fusion (KinectFusion, voxel hashing, Voxblox, nvblox, VDBFusion).",[11],false,"confirmed","author copy","Author-hosted PDF of the SIGGRAPH '96 paper (graphics.stanford.edu\u002Fpapers\u002Fvolrange\u002Fvolrange.pdf, 10 pp.); ACM version of record not compared",[61,67,72],{"category":62,"model":63,"canonical":63,"role":64,"dataset":38,"specs":65,"locator":66},"other","Cyberware 3030 MS","method input","laser stripe optical triangulation scanner; laser sheet from a cylindrical lens with an off-axis CCD; object translates through the laser plane; modified to permit spacetime triangulation","Sec. 5.1, Fig. 1(b), Fig. 12 caption",{"category":68,"model":69,"canonical":69,"role":64,"dataset":38,"specs":70,"locator":71},"platform","motion control platform (model not reported)","used to reposition objects automatically between scans","Sec. 6",{"category":73,"model":74,"canonical":74,"role":75,"dataset":38,"specs":76,"locator":71},"compute","MIPS R4400","compute for runtime","250 MHz processor",[],{"totalRows":79,"groupCount":80,"groups":81,"others":443},30,5,[82,166,275,350],{"slug":83,"group":84,"sourceId":5,"sourceLabel":6,"table":85,"selfRows":86,"metrics":87,"seqs":98,"entrants":118,"cells":128,"outcomes":159,"locators":160,"hardware":161,"wordings":163,"notes":164},"curless1996volumetric-fig-8","curless1996volumetric:Fig. 8","Fig. 8",12,[88,92,95],{"label":89,"unit":90,"statistic":91,"alignment":91},"Exec. time (min)","min","not_reported",{"label":93,"unit":94,"statistic":91,"alignment":91},"Holes (count)","count",{"label":96,"unit":97,"statistic":91,"alignment":91},"Output triangles","million triangles",[99,103,105,108,110,112,114,116],{"dataset":100,"sequence":101,"environment":102},"Dragon (authors' scans)","61 scans, 15 M input triangles","tabletop object",{"dataset":100,"sequence":104,"environment":102},"71 scans incl. backdrop, 24 M input triangles",{"dataset":106,"sequence":107,"environment":102},"Happy Buddha (authors' scans)","48 scans, 5 M input triangles",{"dataset":106,"sequence":109,"environment":102},"58 scans incl. backdrop, 9 M input triangles",{"dataset":100,"sequence":111,"environment":102},"61 scans",{"dataset":100,"sequence":113,"environment":102},"71 scans incl. backdrop",{"dataset":106,"sequence":115,"environment":102},"48 scans",{"dataset":106,"sequence":117,"environment":102},"58 scans incl. backdrop",[119,122,124,126],{"name":120,"methodId":5,"linkable":121,"proposed":121,"self":121},"Dragon (volumetric integration, no hole filling)",true,{"name":123,"methodId":5,"linkable":121,"proposed":121,"self":121},"Dragon + fill (space carving and hole filling)",{"name":125,"methodId":5,"linkable":121,"proposed":121,"self":121},"Buddha (volumetric integration, no hole filling)",{"name":127,"methodId":5,"linkable":121,"proposed":121,"self":121},"Buddha + fill (space carving and hole filling)",[129,133,136,139,142,145,146,149,151,153,155,157],[130,130,130,131,132,130,130,132,130],0,56,-1,[134,130,134,135,132,130,130,132,130],1,257,[137,130,137,138,132,130,130,132,130],2,47,[140,130,140,141,132,130,130,132,130],3,197,[130,134,143,144,132,130,132,132,130],4,324,[134,134,80,130,132,130,132,132,130],[137,134,147,148,132,130,132,132,130],6,670,[140,134,150,130,132,130,132,132,130],7,[130,137,143,152,132,130,132,132,130],1.7,[134,137,80,154,132,130,132,132,130],1.8,[137,137,147,156,132,130,132,132,130],2.4,[140,137,150,158,132,130,132,132,130],2.6,[],[85],[162],"250 MHz MIPS R4400",[],[165],"Reconstruction statistics with and without space carving and hole filling; Dragon voxel 0.35 mm (712x501x322), Buddha voxel 0.25 mm (407x957x407)",{"slug":167,"group":168,"sourceId":169,"sourceLabel":170,"table":171,"selfRows":172,"metrics":173,"seqs":185,"entrants":192,"cells":205,"outcomes":269,"locators":270,"hardware":271,"wordings":272,"notes":273},"coslam2023-table-1","coslam2023:Table 1","coslam2023","Wang et al., 2023a","Table 1",8,[174,178,180,182],{"label":175,"unit":176,"statistic":177,"alignment":32},"Depth L1 (cm)","cm","mean",{"label":179,"unit":176,"statistic":177,"alignment":32},"Acc. (cm)",{"label":181,"unit":176,"statistic":177,"alignment":32},"Comp. (cm)",{"label":183,"unit":184,"statistic":177,"alignment":32},"Comp. Ratio (%, \u003C 5 cm)","%",[186,190],{"dataset":187,"sequence":188,"environment":189},"Replica (8 synthetic scenes)","average over scenes","synthetic indoor",{"dataset":191,"sequence":188,"environment":189},"Synthetic RGB-D of NeuralRGBD (7 scenes, simulated depth noise)",[193,195,198,201,203],{"name":194,"methodId":5,"linkable":121,"proposed":56,"self":121},"TSDF-Fusion",{"name":196,"methodId":197,"linkable":121,"proposed":56,"self":56},"iMAP","imap2021",{"name":199,"methodId":200,"linkable":121,"proposed":56,"self":56},"NICE-SLAM","niceslam2022",{"name":202,"methodId":169,"linkable":121,"proposed":121,"self":56},"Co-SLAM (Ours)",{"name":204,"methodId":197,"linkable":121,"proposed":56,"self":56},"iMAP*",[206,208,210,212,214,216,218,220,222,224,226,228,230,232,234,236,238,240,241,243,245,247,249,251,253,255,257,259,261,263,265,267],[130,130,130,207,132,130,132,132,130],6.36,[130,134,130,209,132,130,132,132,130],1.62,[130,137,130,211,132,130,132,132,130],3.94,[130,140,130,213,132,130,132,132,130],83.93,[134,130,130,215,132,130,132,132,130],4.64,[134,134,130,217,132,130,132,132,130],3.62,[134,137,130,219,132,130,132,132,130],4.93,[134,140,130,221,132,130,132,132,130],80.51,[137,130,130,223,132,130,132,132,130],1.9,[137,134,130,225,132,130,132,132,130],2.37,[137,137,130,227,132,130,132,132,130],2.64,[137,140,130,229,132,130,132,132,130],91.13,[140,130,130,231,132,130,132,132,130],1.51,[140,134,130,233,132,130,132,132,130],2.1,[140,137,130,235,132,130,132,132,130],2.08,[140,140,130,237,132,130,132,132,130],93.44,[130,130,134,239,132,130,132,132,130],10.87,[130,134,134,209,132,130,132,132,130],[130,137,134,242,132,130,132,132,130],5.16,[130,140,134,244,132,130,132,132,130],81.52,[143,130,134,246,132,130,132,132,130],43.91,[143,134,134,248,132,130,132,132,130],18.3,[143,137,134,250,132,130,132,132,130],26.41,[143,140,134,252,132,130,132,132,130],20.73,[137,130,134,254,132,130,132,132,130],6.32,[137,134,134,256,132,130,132,132,130],5.96,[137,137,134,258,132,130,132,132,130],5.3,[137,140,134,260,132,130,132,132,130],77.46,[140,130,134,262,132,130,132,132,130],3.02,[140,134,134,264,132,130,132,132,130],2.95,[140,137,134,266,132,130,132,132,130],2.96,[140,140,134,268,132,130,132,132,130],86.88,[],[171],[],[],[274],"Averages over scenes; meshes culled with the authors' new culling strategy for all methods; TSDF-Fusion uses Co-SLAM poses; iMAP* is the NICE-SLAM re-implementation",{"slug":276,"group":277,"sourceId":200,"sourceLabel":278,"table":171,"selfRows":80,"metrics":279,"seqs":291,"entrants":296,"cells":304,"outcomes":344,"locators":345,"hardware":346,"wordings":347,"notes":348},"niceslam2022-table-1","niceslam2022:Table 1","Zhu et al., 2022a",[280,283,285,287,289],{"label":281,"unit":282,"statistic":91,"alignment":91},"Mem. (MB)","MB",{"label":284,"unit":176,"statistic":177,"alignment":91},"Depth L1 (L1 on 1000 depth maps rendered from meshes)",{"label":286,"unit":176,"statistic":177,"alignment":91},"Acc. [cm]",{"label":288,"unit":176,"statistic":177,"alignment":91},"Comp. [cm]",{"label":290,"unit":184,"statistic":91,"alignment":91},"Comp. Ratio [\u003C 5cm %]",[292],{"dataset":293,"sequence":294,"environment":295},"Replica","average of 8 scenes","synthetic indoor rooms and offices",[297,299,301,303],{"name":298,"methodId":5,"linkable":121,"proposed":56,"self":121},"TSDF-Fusion [11] (with NICE-SLAM poses)",{"name":300,"methodId":197,"linkable":121,"proposed":56,"self":56},"iMAP* [47] (re-implementation)",{"name":302,"methodId":38,"linkable":56,"proposed":56,"self":56},"DI-Fusion [16]",{"name":199,"methodId":200,"linkable":121,"proposed":121,"self":56},[305,307,309,311,313,315,317,319,321,323,325,327,329,331,333,335,337,339,341,342],[130,130,130,306,132,130,132,132,130],67.1,[130,134,130,308,132,130,132,132,130],7.57,[130,137,130,310,132,130,132,132,130],1.6,[130,140,130,312,132,130,132,132,130],3.49,[130,143,130,314,132,130,132,132,130],86.08,[134,130,130,316,132,130,132,132,130],1.04,[134,134,130,318,132,130,132,132,130],7.64,[134,137,130,320,132,130,132,132,130],6.95,[134,140,130,322,132,130,132,132,130],5.33,[134,143,130,324,132,130,132,132,130],66.6,[137,130,130,326,132,130,132,132,130],3.78,[137,134,130,328,132,130,132,132,130],23.33,[137,137,130,330,132,130,132,132,130],19.4,[137,140,130,332,132,130,132,132,130],10.19,[137,143,130,334,132,130,132,132,130],72.96,[140,130,130,336,132,130,132,132,130],12.02,[140,134,130,338,132,130,132,132,130],3.53,[140,137,130,340,132,130,132,132,130],2.85,[140,140,130,140,132,130,132,132,130],[140,143,130,343,132,130,132,132,130],89.33,[],[171],[],[],[349],"Replica, average over 8 scenes and 5 runs; 3D metrics computed after removing regions outside every camera frustum; TSDF-Fusion uses NICE-SLAM poses at 256^3 voxels; iMAP* is the authors' re-implementation",{"slug":351,"group":352,"sourceId":353,"sourceLabel":354,"table":355,"selfRows":140,"metrics":356,"seqs":366,"entrants":371,"cells":388,"outcomes":437,"locators":438,"hardware":439,"wordings":440,"notes":441},"kazhdan2006poisson-table-2","kazhdan2006poisson:Table 2","kazhdan2006poisson","Kazhdan et al., 2006","Table 2",[357,360,363],{"label":358,"unit":359,"statistic":91,"alignment":91},"Time (s)","s",{"label":361,"unit":282,"statistic":362,"alignment":91},"Peak Memory (MB)","max",{"label":364,"unit":365,"statistic":91,"alignment":91},"# of Tris.","triangles",[367],{"dataset":368,"sequence":369,"environment":370},"Stanford Bunny","raw data, 362,000 points (Poisson at depth 9)","object scan",[372,374,376,378,380,382,384,386],{"name":373,"methodId":38,"linkable":56,"proposed":56,"self":56},"Power Crust",{"name":375,"methodId":38,"linkable":56,"proposed":56,"self":56},"Robust Cocone",{"name":377,"methodId":38,"linkable":56,"proposed":56,"self":56},"FastRBF",{"name":379,"methodId":38,"linkable":56,"proposed":56,"self":56},"MPU",{"name":381,"methodId":38,"linkable":56,"proposed":56,"self":56},"Hoppe et al 1992",{"name":383,"methodId":5,"linkable":121,"proposed":56,"self":121},"VRIP",{"name":385,"methodId":38,"linkable":56,"proposed":56,"self":56},"FFT",{"name":387,"methodId":353,"linkable":121,"proposed":121,"self":56},"Poisson",[389,391,393,395,397,399,401,403,405,407,409,411,413,415,417,419,421,423,425,427,429,431,433,435],[130,130,130,390,132,130,132,132,130],380,[134,130,130,392,132,130,132,132,130],892,[137,130,130,394,132,130,132,132,130],4919,[140,130,130,396,132,130,132,132,130],28,[143,130,130,398,132,130,132,132,130],70,[80,130,130,400,132,130,132,132,130],86,[147,130,130,402,132,130,132,132,130],125,[150,130,130,404,132,130,132,132,130],263,[130,134,130,406,132,130,132,132,130],2653,[134,134,130,408,132,130,132,132,130],544,[137,134,130,410,132,130,132,132,130],796,[140,134,130,412,132,130,132,132,130],260,[143,134,130,414,132,130,132,132,130],330,[80,134,130,416,132,130,132,132,130],186,[147,134,130,418,132,130,132,132,130],1684,[150,134,130,420,132,130,132,132,130],310,[130,137,130,422,132,130,132,132,130],554332,[134,137,130,424,132,130,132,132,130],272662,[137,137,130,426,132,130,132,132,130],1798154,[140,137,130,428,132,130,132,132,130],925240,[143,137,130,430,132,130,132,132,130],950562,[80,137,130,432,132,130,132,132,130],1038055,[147,137,130,434,132,130,132,132,130],910320,[150,137,130,436,132,130,132,132,130],911390,[],[355],[],[],[442],"Stanford Bunny raw data (362,000 points from ten range images), processed to fit each algorithm's input format; Poisson reconstructed at octree depth 9, resolution settings of the other seven methods not reported (VRIP used the registered scans with confidence values); running time in seconds, peak memory in MB, output triangles; hardware not reported",[444],{"group":445,"slug":446,"sourceLabel":6,"table":447,"selfRows":137,"datasets":448},"curless1996volumetric:Text Sec. 6","curless1996volumetric-text-sec-6","Text Sec. 6",[100,106],1790510664550]