[{"data":1,"prerenderedAt":581},["ShallowReactive",2],{"method-fischler1981ransac":3},{"method":4,"reference":46,"equipment":65,"figures":72,"results":73},{"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":27,"sensors":32,"platform":35,"estimator":37,"association":38,"timeModel":39,"deskew":39,"loopClosure":40,"globalOptimization":40,"mapRepresentation":39,"prior":40,"outputGeometry":41,"compute":42,"codeUrl":43,"codeLicense":44,"relatedVersions":45},"fischler1981ransac","Fischler & Bolles, 1981","RANSAC","Random sample consensus",1981,"classic","C02","registration_component","隨機取樣一致（RANSAC）以最少數量的資料點實例化模型，再收集誤差容許範圍內的一致集合；若一致集合大小達門檻 t，就在該集合上以最小平方法重新估計，否則重新抽樣，試驗次數用盡時採用最大一致集合或宣告失敗（Sec. II）。論文推導試驗次數期望值 E(k) = w^(-n) 與達成信心 z 所需的 k。主要應用為由已知位置地標的影像求相機投影中心的定位問題（LDP），給出 P3P 最多四解的封閉形式解，並證明共面 P4P 與一般位置 P6P 有唯一解；原文不涉及點雲配準。點雲配準中的用法依 FGR 論文 Sec. 1 的描述，TEASER 論文 Sec. I 註腳指出其執行時間隨離群比例呈指數成長。","Hypothesize-and-verify robust fitting originally applied to the Location Determination Problem; in registration it is the standard coarse-alignment baseline whose runtime grows exponentially with the outlier ratio (per yang2021teaser Sec. I, footnote 1).","full_text_reviewed","peer_reviewed_published","background","not_reported",[20],"simulation",[22,23,24,25,26],"interprets and smooths data containing a significant percentage of gross errors (abstract)","in an LDP with 5 gross errors among 20 correspondences a least-squares pruning heuristic kept 3 gross errors, while RANSAC found the correct solution on the second triple with none (Sec. IV.C)","in 50 synthetic LDPs no gross error entered the final consensus set (Sec. IV.D, Table I)","real aerial image: consensus of 17 on the first triple, extended to all 22 good correspondences (Sec. IV.E)","closed-form P3P solution with up to four real solutions; unique solutions for coplanar P4P and general-position P6P (Sec. III, Appendices A-B)",[28,29,30,31],"runtime grows exponentially with outlier ratio and performs poorly at high outlier rates (yang2021teaser Sec. I, secondary)","three parameters must be set: error tolerance, number of trials and consensus threshold t (Sec. II)","expected trials rise as w^(-n), for example 16 for w = 0.5 and n = 4 but 625 for w = 0.2 and n = 4, and SD(k) is about equal to E(k), so two to three times E(k) trials may be needed (Sec. II.B)","multiple physically real solutions can exist for P3P, P4P and P5P, so a sample may yield ambiguous poses (Sec. III)",[33,34],"aerial photograph from about 4,000 ft with a 6 in. lens, digitized on a 2,000 x 2,000 grid (about 2 ft per pixel)","synthetic landmark-to-image correspondences",[20,36],"airborne camera (aerial image)","hypothesize-and-verify: randomly select a minimal subset of n data points to instantiate the model, collect the consensus set within an error tolerance, and if its size reaches threshold t refit the model (e.g., least squares) on the consensus set; otherwise resample; after k trials use the largest consensus set or fail; expected trials E(k) = w^(-n), and k = log(1-z)\u002Flog(1-w^n) for confidence z","original: landmark-to-image correspondences from error-prone feature detectors (cross correlation in the aerial test); consensus judged with image-plane error ellipses derived from perturbing the three selected points; point cloud registration use with putative 3D correspondences is a later application (secondary)","not_applicable","none","in the original LDP application: 3-D location of the centre of perspective with an error estimate, and the spatial orientation of the image plane","about 1 s per camera position considered for the synthetic LDP program; hardware not reported",null,"not_verified",[],{"id":5,"kind":47,"shortName":7,"title":8,"authors":48,"year":9,"venue":51,"venueType":52,"publisher":53,"volumeIssuePages":54,"doi":55,"arxivId":43,"url":56,"firstPublicDate":57,"publicationStatus":16,"metadataStatus":58,"fulltextStatus":15,"era":10,"classicReason":59,"codeUrl":43,"cluster":11,"topics":60,"mdpi":61,"verification":62,"label":6,"fulltextRoute":63,"versionRead":64,"addedByCensus":61},"component",[49,50],"Martin A. Fischler","Robert C. Bolles","Communications of the ACM","journal","ACM","24(6):381-395","10.1145\u002F358669.358692","https:\u002F\u002Fapi.crossref.org\u002Fworks\u002F10.1145\u002F358669.358692","1981-06","metadata_verified","reproducible baseline: RANSAC is the hypothesize-and-verify baseline against which FGR, TEASER++ and KISS-Matcher report comparisons (zhou2016fgr Sec. 1; yang2021teaser Sec. XI).",[11],false,"corrected","publisher OA","ACM Digital Library version of record, Communications of the ACM 24(6):381-395 (June 1981), 15-page scanned PDF with text layer (all pages incl. Appendices A and B read)",[66],{"category":67,"model":68,"canonical":68,"role":69,"dataset":43,"specs":70,"locator":71},"camera","aerial camera with 6 in. lens","method input","image taken from approximately 4,000 ft; digitized on a 2,000 x 2,000 pixel grid, about 2 ft per pixel ground resolution","Sec. IV.E",[],{"totalRows":74,"groupCount":75,"groups":76,"others":538},75,11,[77,174,296,348],{"slug":78,"group":79,"sourceId":5,"sourceLabel":6,"table":80,"selfRows":81,"metrics":82,"seqs":90,"entrants":112,"cells":116,"outcomes":168,"locators":169,"hardware":170,"wordings":171,"notes":172},"fischler1981ransac-table-i","fischler1981ransac:Table I","Table I",30,[83,86,88],{"label":84,"unit":85,"statistic":18,"alignment":40},"No. of correspondences in final consensus set","count",{"label":87,"unit":85,"statistic":18,"alignment":40},"No. of triples considered",{"label":89,"unit":85,"statistic":18,"alignment":40},"No. of camera positions considered",[91,94,96,98,100,102,104,106,108,110],{"dataset":92,"sequence":93,"environment":20},"50 synthetic location determination problems","typical problem 1 (w=0.8)",{"dataset":92,"sequence":95,"environment":20},"typical problem 2 (w=0.8)",{"dataset":92,"sequence":97,"environment":20},"typical problem 3 (w=0.8)",{"dataset":92,"sequence":99,"environment":20},"typical problem 4 (w=0.8)",{"dataset":92,"sequence":101,"environment":20},"typical problem 5 (w=0.8)",{"dataset":92,"sequence":103,"environment":20},"typical problem 1 (w=0.6)",{"dataset":92,"sequence":105,"environment":20},"typical problem 2 (w=0.6)",{"dataset":92,"sequence":107,"environment":20},"typical problem 3 (w=0.6)",{"dataset":92,"sequence":109,"environment":20},"typical problem 4 (w=0.6)",{"dataset":92,"sequence":111,"environment":20},"typical problem 5 (w=0.6)",[113],{"name":114,"methodId":5,"linkable":115,"proposed":115,"self":115},"RANSAC\u002FLD",true,[117,121,124,127,129,130,132,133,134,135,137,138,139,141,142,144,147,148,150,152,153,154,157,158,159,160,162,163,165,166],[118,118,118,119,120,118,120,120,118],0,19,-1,[118,122,118,123,120,118,120,120,118],1,6,[118,125,118,126,120,118,120,120,118],2,10,[118,118,122,128,120,118,120,120,118],23,[118,122,122,122,120,118,120,120,118],[118,125,122,131,120,118,120,120,118],3,[118,118,125,119,120,118,120,120,118],[118,122,125,125,120,118,120,120,118],[118,125,125,131,120,118,120,120,118],[118,118,131,136,120,118,120,120,118],25,[118,122,131,122,120,118,120,120,118],[118,125,131,125,120,118,120,120,118],[118,118,140,128,120,118,120,120,118],4,[118,122,140,131,120,118,120,120,118],[118,125,140,143,120,118,120,120,118],8,[118,118,145,146,120,118,120,120,118],5,20,[118,122,145,75,120,118,120,120,118],[118,125,145,149,120,118,120,120,118],21,[118,118,123,151,120,118,120,120,118],17,[118,122,123,122,120,118,120,120,118],[118,125,123,122,120,118,120,120,118],[118,118,155,156,120,118,120,120,118],7,16,[118,122,155,123,120,118,120,120,118],[118,125,155,143,120,118,120,120,118],[118,118,143,156,120,118,120,120,118],[118,122,143,161,120,118,120,120,118],9,[118,125,143,149,120,118,120,120,118],[118,118,161,164,120,118,120,120,118],18,[118,122,161,161,120,118,120,120,118],[118,125,161,167,120,118,120,120,118],15,[],[80],[],[],[173],"Ten typical of 50 synthetic LDPs, 30 landmark-to-image correspondences each; gross errors at least 10 px off, good ones with 1 px std; RANSAC avoided gross errors in the final consensus set in all problems",{"slug":175,"group":176,"sourceId":177,"sourceLabel":178,"table":179,"selfRows":164,"metrics":180,"seqs":188,"entrants":209,"cells":218,"outcomes":288,"locators":289,"hardware":290,"wordings":292,"notes":293},"yang2021teaser-table-ii","yang2021teaser:Table II","yang2021teaser","Yang et al., 2021","Table II",[181,184],{"label":182,"unit":183,"statistic":18,"alignment":40},"percentage of correct registration results","%",{"label":185,"unit":186,"statistic":187,"alignment":40},"Avg. Runtime [s]","s","mean",[189,193,195,197,199,201,203,205,207],{"dataset":190,"sequence":191,"environment":192},"3DMatch","Kitchen","indoor RGB-D scans",{"dataset":190,"sequence":194,"environment":192},"Home 1",{"dataset":190,"sequence":196,"environment":192},"Home 2",{"dataset":190,"sequence":198,"environment":192},"Hotel 1",{"dataset":190,"sequence":200,"environment":192},"Hotel 2",{"dataset":190,"sequence":202,"environment":192},"Hotel 3",{"dataset":190,"sequence":204,"environment":192},"Study",{"dataset":190,"sequence":206,"environment":192},"MIT Lab",{"dataset":190,"sequence":208,"environment":192},"all 8 test scenes",[210,212,214,216],{"name":211,"methodId":5,"linkable":115,"proposed":61,"self":115},"RANSAC-1K",{"name":213,"methodId":5,"linkable":115,"proposed":61,"self":115},"RANSAC-10K",{"name":215,"methodId":177,"linkable":115,"proposed":115,"self":61},"TEASER++",{"name":217,"methodId":177,"linkable":115,"proposed":115,"self":61},"TEASER++ (CERT)",[219,221,223,225,227,229,231,233,235,237,239,241,243,245,247,249,251,253,255,257,259,260,262,264,265,267,269,271,273,275,277,279,281,282,284,286],[118,118,118,220,120,118,120,120,118],91.3,[118,118,122,222,120,118,120,120,122],89.1,[118,118,125,224,120,118,120,120,122],74.5,[118,118,131,226,120,118,120,120,122],94.2,[118,118,140,228,120,118,120,120,122],84.6,[118,118,145,230,120,118,120,120,122],90.7,[118,118,123,232,120,118,120,120,122],86.3,[118,118,155,234,120,118,120,120,122],81.8,[118,122,143,236,120,118,118,120,122],0.008,[122,118,118,238,120,118,120,120,122],97.2,[122,118,122,240,120,118,120,120,122],92.3,[122,118,125,242,120,118,120,120,122],79.3,[122,118,131,244,120,118,120,120,122],96.5,[122,118,140,246,120,118,120,120,122],86.5,[122,118,145,248,120,118,120,120,122],94.4,[122,118,123,250,120,118,120,120,122],90.4,[122,118,155,252,120,118,120,120,122],85.7,[122,122,143,254,120,118,118,120,122],0.074,[125,118,118,256,120,118,120,120,122],98.6,[125,118,122,258,120,118,120,120,122],92.9,[125,118,125,246,120,118,120,120,122],[125,118,131,261,120,118,120,120,122],97.8,[125,118,140,263,120,118,120,120,122],89.4,[125,118,145,248,120,118,120,120,122],[125,118,123,266,120,118,120,120,122],91.1,[125,118,155,268,120,118,120,120,122],83.1,[125,122,143,270,120,118,118,120,122],0.059,[131,118,118,272,120,118,120,120,122],99.4,[131,118,122,274,120,118,120,120,122],94.1,[131,118,125,276,120,118,120,120,122],88.7,[131,118,131,278,120,118,120,120,122],98.2,[131,118,140,280,120,118,120,120,122],91.9,[131,118,145,248,120,118,120,120,122],[131,118,123,283,120,118,120,120,122],94.3,[131,118,155,285,120,118,120,120,122],88.6,[131,122,143,287,120,118,118,120,122],238.136,[],[179],[291],"Xeon Platinum 8259CL at 2.50 GHz, 12 threads",[],[294,295],"3DMatch test scenes, 3DSmoothNet correspondences, success = rotation error \u003C 10 deg and translation error \u003C 30 cm; beta = 5 cm; CERT = subset certified with sub-optimality gap 3%; Xeon Platinum 8259CL, 12 threads","Same setting as other Table II rows",{"slug":297,"group":298,"sourceId":5,"sourceLabel":6,"table":299,"selfRows":143,"metrics":300,"seqs":320,"entrants":325,"cells":327,"outcomes":342,"locators":343,"hardware":344,"wordings":345,"notes":346},"fischler1981ransac-text-sec-iv-e","fischler1981ransac:Text Sec. IV.E","Text Sec. IV.E",[301,303,305,309,311,313,316,318],{"label":302,"unit":85,"statistic":18,"alignment":40},"consensus set size found on the first triple",{"label":304,"unit":85,"statistic":18,"alignment":40},"final consensus set after initial least-squares fit (all good correspondences)",{"label":306,"unit":307,"statistic":308,"alignment":40},"final standard deviation of camera parameter X","ft","std",{"label":310,"unit":307,"statistic":308,"alignment":40},"final standard deviation of camera parameter Y",{"label":312,"unit":307,"statistic":308,"alignment":40},"final standard deviation of camera parameter Z",{"label":314,"unit":315,"statistic":308,"alignment":40},"final standard deviation of camera parameter Heading","deg",{"label":317,"unit":315,"statistic":308,"alignment":40},"final standard deviation of camera parameter Pitch",{"label":319,"unit":315,"statistic":308,"alignment":40},"final standard deviation of camera parameter Roll",[321],{"dataset":322,"sequence":323,"environment":324},"real aerial image","single image","aerial imagery",[326],{"name":114,"methodId":5,"linkable":115,"proposed":115,"self":115},[328,329,331,333,335,337,339,340],[118,118,118,151,120,118,120,120,118],[118,122,118,330,120,118,120,120,118],22,[118,125,118,332,120,118,120,120,118],0.1,[118,131,118,334,120,118,120,120,118],6.4,[118,140,118,336,120,118,120,120,118],2.1,[118,145,118,338,120,118,120,120,118],0.01,[118,123,118,332,120,118,120,120,118],[118,155,118,341,120,118,120,120,118],0.12,[],[71],[],[],[347],"Aerial image from about 4,000 ft with 6 in. lens, digitized 2,000 x 2,000 px (about 2 ft per px); 25 landmarks found by cross correlation, 3 gross errors",{"slug":349,"group":350,"sourceId":351,"sourceLabel":352,"table":353,"selfRows":123,"metrics":354,"seqs":369,"entrants":374,"cells":399,"outcomes":532,"locators":533,"hardware":534,"wordings":535,"notes":536},"zhang2024globalbimreg-table-1","zhang2024globalbimreg:Table 1","zhang2024globalbimreg","Zhang et al., 2024b","Table 1",[355,358,360,362,365,367],{"label":356,"unit":315,"statistic":357,"alignment":40},"RE_50, rotation error 50th percentile","median",{"label":359,"unit":315,"statistic":18,"alignment":40},"RE_75, rotation error 75th percentile",{"label":361,"unit":315,"statistic":18,"alignment":40},"RE_95, rotation error 95th percentile",{"label":363,"unit":364,"statistic":357,"alignment":40},"TE_50, translation error 50th percentile","m",{"label":366,"unit":364,"statistic":18,"alignment":40},"TE_75, translation error 75th percentile",{"label":368,"unit":364,"statistic":18,"alignment":40},"TE_95, translation error 95th percentile",[370],{"dataset":371,"sequence":372,"environment":373},"ISPRS benchmark on indoor modelling","Models 01-05 (250 samples)","building interiors from the ISPRS indoor modelling benchmark (Models 01-05)",[375,377,379,382,384,387,389,391,393,395,397],{"name":376,"methodId":43,"linkable":61,"proposed":61,"self":61},"GMMTree (initialised with FPFH-RANSAC)",{"name":378,"methodId":43,"linkable":61,"proposed":61,"self":61},"FilterReg (initialised with FPFH-RANSAC)",{"name":380,"methodId":381,"linkable":115,"proposed":61,"self":61},"GO-ICP","yang2016goicp",{"name":383,"methodId":43,"linkable":61,"proposed":61,"self":61},"Super4PCS",{"name":385,"methodId":386,"linkable":115,"proposed":61,"self":61},"FGR","zhou2016fgr",{"name":388,"methodId":5,"linkable":115,"proposed":61,"self":115},"RANSAC (FPFH features)",{"name":390,"methodId":43,"linkable":61,"proposed":61,"self":61},"RMMG",{"name":392,"methodId":43,"linkable":61,"proposed":61,"self":61},"PLADE",{"name":394,"methodId":43,"linkable":61,"proposed":61,"self":61},"DCP",{"name":396,"methodId":43,"linkable":61,"proposed":61,"self":61},"PointNetLK",{"name":398,"methodId":351,"linkable":115,"proposed":115,"self":61},"Ours (primitive-level coarse registration)",[400,402,404,406,408,410,412,414,416,418,420,422,424,426,428,430,432,434,436,438,440,442,444,446,448,450,452,454,456,458,460,462,464,466,468,470,472,474,476,478,480,482,484,486,488,490,492,494,496,498,500,502,504,506,508,510,512,514,516,518,520,522,524,526,528,530],[118,118,118,401,120,118,120,120,118],4.466,[118,122,118,403,120,118,120,120,118],8.214,[118,125,118,405,120,118,120,120,118],19.348,[118,131,118,407,120,118,120,120,118],1.652,[118,140,118,409,120,118,120,120,118],3.166,[118,145,118,411,120,118,120,120,118],5.419,[122,118,118,413,120,118,120,120,118],3.59,[122,122,118,415,120,118,120,120,118],10.906,[122,125,118,417,120,118,120,120,118],18.869,[122,131,118,419,120,118,120,120,118],1.076,[122,140,118,421,120,118,120,120,118],1.768,[122,145,118,423,120,118,120,120,118],5.279,[125,118,118,425,120,118,120,120,118],4.178,[125,122,118,427,120,118,120,120,118],4.252,[125,125,118,429,120,118,120,120,118],10.618,[125,131,118,431,120,118,120,120,118],1.787,[125,140,118,433,120,118,120,120,118],2.398,[125,145,118,435,120,118,120,120,118],5.216,[131,118,118,437,120,118,120,120,118],2.208,[131,122,118,439,120,118,120,120,118],5.162,[131,125,118,441,120,118,120,120,118],10.693,[131,131,118,443,120,118,120,120,118],1.14,[131,140,118,445,120,118,120,120,118],2.31,[131,145,118,447,120,118,120,120,118],5.591,[140,118,118,449,120,118,120,120,118],27.139,[140,122,118,451,120,118,120,120,118],35.195,[140,125,118,453,120,118,120,120,118],44.235,[140,131,118,455,120,118,120,120,118],1.969,[140,140,118,457,120,118,120,120,118],3.674,[140,145,118,459,120,118,120,120,118],5.094,[145,118,118,461,120,118,120,120,118],5.477,[145,122,118,463,120,118,120,120,118],7.912,[145,125,118,465,120,118,120,120,118],13.7,[145,131,118,467,120,118,120,120,118],1.412,[145,140,118,469,120,118,120,120,118],2.032,[145,145,118,471,120,118,120,120,118],5.253,[123,118,118,473,120,118,120,120,118],1.673,[123,122,118,475,120,118,120,120,118],3.351,[123,125,118,477,120,118,120,120,118],5.669,[123,131,118,479,120,118,120,120,118],1.335,[123,140,118,481,120,118,120,120,118],1.983,[123,145,118,483,120,118,120,120,118],8.136,[155,118,118,485,120,118,120,120,118],0.424,[155,122,118,487,120,118,120,120,118],0.623,[155,125,118,489,120,118,120,120,118],1.473,[155,131,118,491,120,118,120,120,118],3.83,[155,140,118,493,120,118,120,120,118],6.554,[155,145,118,495,120,118,120,120,118],9.659,[143,118,118,497,120,118,120,120,118],34.865,[143,122,118,499,120,118,120,120,118],43.069,[143,125,118,501,120,118,120,120,118],44.88,[143,131,118,503,120,118,120,120,118],1.945,[143,140,118,505,120,118,120,120,118],2.667,[143,145,118,507,120,118,120,120,118],9.945,[161,118,118,509,120,118,120,120,118],11.741,[161,122,118,511,120,118,120,120,118],21.618,[161,125,118,513,120,118,120,120,118],40.482,[161,131,118,515,120,118,120,120,118],5.007,[161,140,118,517,120,118,120,120,118],6.566,[161,145,118,519,120,118,120,120,118],9.231,[126,118,118,521,120,118,120,120,118],0.272,[126,122,118,523,120,118,120,120,118],0.329,[126,125,118,525,120,118,120,120,118],0.409,[126,131,118,527,120,118,120,120,118],0.053,[126,140,118,529,120,118,120,120,118],0.148,[126,145,118,531,120,118,120,120,118],0.352,[],[353],[],[],[537],"Coarse registration on 250 samples (50 per model) with random rigid perturbations (roll and pitch within 30 deg, yaw within 180 deg, translation within 10 m); only successful results with RE \u003C 45 deg and TE \u003C 10 m are included; A50, A75 and A95 quantiles of rotation and translation error against the benchmark alignment",[539,546,552,559,564,570,576],{"group":540,"slug":541,"sourceLabel":542,"table":543,"selfRows":145,"datasets":544},"qin2023geotransformer:Table 5","qin2023geotransformer-table-5","Qin et al., 2023","Table 5",[545],"Augmented ICL-NUIM",{"group":547,"slug":548,"sourceLabel":6,"table":549,"selfRows":125,"datasets":550},"fischler1981ransac:Text Sec. IV.C","fischler1981ransac-text-sec-iv-c","Text Sec. IV.C",[551],"synthetic LDP (20 landmarks)",{"group":553,"slug":554,"sourceLabel":555,"table":556,"selfRows":125,"datasets":557},"yang2020gnc:Text Sec.V-A P-REG","yang2020gnc-text-sec-v-a-p-reg","Yang et al., 2020b","Text Sec.V-A P-REG",[558],"Stanford 3D Scanning Repository Bunny",{"group":560,"slug":561,"sourceLabel":6,"table":562,"selfRows":122,"datasets":563},"fischler1981ransac:Text Sec. IV.D","fischler1981ransac-text-sec-iv-d","Text Sec. IV.D",[92],{"group":565,"slug":566,"sourceLabel":555,"table":567,"selfRows":122,"datasets":568},"yang2020gnc:Text Sec.V-A G-REG","yang2020gnc-text-sec-v-a-g-reg","Text Sec.V-A G-REG",[569],"PASCAL+ (car-2 mesh)",{"group":571,"slug":572,"sourceLabel":555,"table":573,"selfRows":122,"datasets":574},"yang2020gnc:Text Sec.V-C","yang2020gnc-text-sec-v-c","Text Sec.V-C",[575],"FG3DCar",{"group":577,"slug":578,"sourceLabel":352,"table":579,"selfRows":122,"datasets":580},"zhang2024globalbimreg:Text Sec.4.1.2","zhang2024globalbimreg-text-sec-4-1-2","Text Sec.4.1.2",[371],1790510664314]