Eintrag weiter verarbeiten
SUPERVISED OUTLIER DETECTION IN LARGE-SCALE MVS POINT CLOUDS FOR 3D CITY MODELING APPLICATIONS
Gespeichert in:
| Zeitschriftentitel: | ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences |
|---|---|
| Personen und Körperschaften: | , , , |
| In: | ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, IV-2, 2018, S. 263-270 |
| Format: | E-Article |
| Sprache: | Englisch |
| veröffentlicht: |
Copernicus GmbH
|
| Schlagwörter: |
| author_facet |
Stucker, C. Richard, A. Wegner, J. D. Schindler, K. Stucker, C. Richard, A. Wegner, J. D. Schindler, K. |
|---|---|
| author |
Stucker, C. Richard, A. Wegner, J. D. Schindler, K. |
| spellingShingle |
Stucker, C. Richard, A. Wegner, J. D. Schindler, K. ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences SUPERVISED OUTLIER DETECTION IN LARGE-SCALE MVS POINT CLOUDS FOR 3D CITY MODELING APPLICATIONS General Medicine |
| author_sort |
stucker, c. |
| spelling |
Stucker, C. Richard, A. Wegner, J. D. Schindler, K. 2194-9050 Copernicus GmbH General Medicine http://dx.doi.org/10.5194/isprs-annals-iv-2-263-2018 <jats:p>Abstract. We propose to use a discriminative classifier for outlier detection in large-scale point clouds of cities generated via multi-view stereo (MVS) from densely acquired images. What makes outlier removal hard are varying distributions of inliers and outliers across a scene. Heuristic outlier removal using a specific feature that encodes point distribution often delivers unsatisfying results. Although most outliers can be identified correctly (high recall), many inliers are erroneously removed (low precision), too. This aggravates object 3D reconstruction due to missing data. We thus propose to discriminatively learn class-specific distributions directly from the data to achieve high precision. We apply a standard Random Forest classifier that infers a binary label (inlier or outlier) for each 3D point in the raw, unfiltered point cloud and test two approaches for training. In the first, non-semantic approach, features are extracted without considering the semantic interpretation of the 3D points. The trained model approximates the average distribution of inliers and outliers across all semantic classes. Second, semantic interpretation is incorporated into the learning process, i.e. we train separate inlieroutlier classifiers per semantic class (building facades, roof, ground, vegetation, fields, and water). Performance of learned filtering is evaluated on several large SfM point clouds of cities. We find that results confirm our underlying assumption that discriminatively learning inlier-outlier distributions does improve precision over global heuristics by up to ≈ 12 percent points. Moreover, semantically informed filtering that models class-specific distributions further improves precision by up to ≈ 10 percent points, being able to remove very isolated building, roof, and water points while preserving inliers on building facades and vegetation. </jats:p> SUPERVISED OUTLIER DETECTION IN LARGE-SCALE MVS POINT CLOUDS FOR 3D CITY MODELING APPLICATIONS ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences |
| doi_str_mv |
10.5194/isprs-annals-iv-2-263-2018 |
| facet_avail |
Online |
| format |
ElectronicArticle |
| fullrecord |
blob:ai-49-aHR0cDovL2R4LmRvaS5vcmcvMTAuNTE5NC9pc3Bycy1hbm5hbHMtaXYtMi0yNjMtMjAxOA |
| id |
ai-49-aHR0cDovL2R4LmRvaS5vcmcvMTAuNTE5NC9pc3Bycy1hbm5hbHMtaXYtMi0yNjMtMjAxOA |
| institution |
DE-D275 DE-Bn3 DE-Brt1 DE-Zwi2 DE-D161 DE-Gla1 DE-Zi4 DE-15 DE-Pl11 DE-Rs1 DE-105 DE-14 DE-Ch1 DE-L229 |
| imprint |
Copernicus GmbH, 2018 |
| imprint_str_mv |
Copernicus GmbH, 2018 |
| issn |
2194-9050 |
| issn_str_mv |
2194-9050 |
| language |
English |
| mega_collection |
Copernicus GmbH (CrossRef) |
| match_str |
stucker2018supervisedoutlierdetectioninlargescalemvspointcloudsfor3dcitymodelingapplications |
| publishDateSort |
2018 |
| publisher |
Copernicus GmbH |
| recordtype |
ai |
| record_format |
ai |
| series |
ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences |
| source_id |
49 |
| title |
SUPERVISED OUTLIER DETECTION IN LARGE-SCALE MVS POINT CLOUDS FOR 3D CITY MODELING APPLICATIONS |
| title_unstemmed |
SUPERVISED OUTLIER DETECTION IN LARGE-SCALE MVS POINT CLOUDS FOR 3D CITY MODELING APPLICATIONS |
| title_full |
SUPERVISED OUTLIER DETECTION IN LARGE-SCALE MVS POINT CLOUDS FOR 3D CITY MODELING APPLICATIONS |
| title_fullStr |
SUPERVISED OUTLIER DETECTION IN LARGE-SCALE MVS POINT CLOUDS FOR 3D CITY MODELING APPLICATIONS |
| title_full_unstemmed |
SUPERVISED OUTLIER DETECTION IN LARGE-SCALE MVS POINT CLOUDS FOR 3D CITY MODELING APPLICATIONS |
| title_short |
SUPERVISED OUTLIER DETECTION IN LARGE-SCALE MVS POINT CLOUDS FOR 3D CITY MODELING APPLICATIONS |
| title_sort |
supervised outlier detection in large-scale mvs point clouds for 3d city modeling applications |
| topic |
General Medicine |
| url |
http://dx.doi.org/10.5194/isprs-annals-iv-2-263-2018 |
| publishDate |
2018 |
| physical |
263-270 |
| description |
<jats:p>Abstract. We propose to use a discriminative classifier for outlier detection in large-scale point clouds of cities generated via multi-view stereo (MVS) from densely acquired images. What makes outlier removal hard are varying distributions of inliers and outliers across a scene. Heuristic outlier removal using a specific feature that encodes point distribution often delivers unsatisfying results. Although most outliers can be identified correctly (high recall), many inliers are erroneously removed (low precision), too. This aggravates object 3D reconstruction due to missing data. We thus propose to discriminatively learn class-specific distributions directly from the data to achieve high precision. We apply a standard Random Forest classifier that infers a binary label (inlier or outlier) for each 3D point in the raw, unfiltered point cloud and test two approaches for training. In the first, non-semantic approach, features are extracted without considering the semantic interpretation of the 3D points. The trained model approximates the average distribution of inliers and outliers across all semantic classes. Second, semantic interpretation is incorporated into the learning process, i.e. we train separate inlieroutlier classifiers per semantic class (building facades, roof, ground, vegetation, fields, and water). Performance of learned filtering is evaluated on several large SfM point clouds of cities. We find that results confirm our underlying assumption that discriminatively learning inlier-outlier distributions does improve precision over global heuristics by up to ≈ 12 percent points. Moreover, semantically informed filtering that models class-specific distributions further improves precision by up to ≈ 10 percent points, being able to remove very isolated building, roof, and water points while preserving inliers on building facades and vegetation.
</jats:p> |
| container_start_page |
263 |
| container_title |
ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences |
| container_volume |
IV-2 |
| format_de105 |
Article, E-Article |
| format_de14 |
Article, E-Article |
| format_de15 |
Article, E-Article |
| format_de520 |
Article, E-Article |
| format_de540 |
Article, E-Article |
| format_dech1 |
Article, E-Article |
| format_ded117 |
Article, E-Article |
| format_degla1 |
E-Article |
| format_del152 |
Buch |
| format_del189 |
Article, E-Article |
| format_dezi4 |
Article |
| format_dezwi2 |
Article, E-Article |
| format_finc |
Article, E-Article |
| format_nrw |
Article, E-Article |
| _version_ |
1792344300523094025 |
| geogr_code |
not assigned |
| last_indexed |
2024-03-01T17:05:24.224Z |
| geogr_code_person |
not assigned |
| openURL |
url_ver=Z39.88-2004&ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fvufind.svn.sourceforge.net%3Agenerator&rft.title=SUPERVISED+OUTLIER+DETECTION+IN+LARGE-SCALE+MVS+POINT+CLOUDS+FOR+3D+CITY+MODELING+APPLICATIONS&rft.date=2018-05-28&genre=article&issn=2194-9050&volume=IV-2&spage=263&epage=270&pages=263-270&jtitle=ISPRS+Annals+of+the+Photogrammetry%2C+Remote+Sensing+and+Spatial+Information+Sciences&atitle=SUPERVISED+OUTLIER+DETECTION+IN+LARGE-SCALE+MVS+POINT+CLOUDS+FOR+3D+CITY+MODELING+APPLICATIONS&aulast=Schindler&aufirst=K.&rft_id=info%3Adoi%2F10.5194%2Fisprs-annals-iv-2-263-2018&rft.language%5B0%5D=eng |
| SOLR | |
| _version_ | 1792344300523094025 |
| author | Stucker, C., Richard, A., Wegner, J. D., Schindler, K. |
| author_facet | Stucker, C., Richard, A., Wegner, J. D., Schindler, K., Stucker, C., Richard, A., Wegner, J. D., Schindler, K. |
| author_sort | stucker, c. |
| container_start_page | 263 |
| container_title | ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences |
| container_volume | IV-2 |
| description | <jats:p>Abstract. We propose to use a discriminative classifier for outlier detection in large-scale point clouds of cities generated via multi-view stereo (MVS) from densely acquired images. What makes outlier removal hard are varying distributions of inliers and outliers across a scene. Heuristic outlier removal using a specific feature that encodes point distribution often delivers unsatisfying results. Although most outliers can be identified correctly (high recall), many inliers are erroneously removed (low precision), too. This aggravates object 3D reconstruction due to missing data. We thus propose to discriminatively learn class-specific distributions directly from the data to achieve high precision. We apply a standard Random Forest classifier that infers a binary label (inlier or outlier) for each 3D point in the raw, unfiltered point cloud and test two approaches for training. In the first, non-semantic approach, features are extracted without considering the semantic interpretation of the 3D points. The trained model approximates the average distribution of inliers and outliers across all semantic classes. Second, semantic interpretation is incorporated into the learning process, i.e. we train separate inlieroutlier classifiers per semantic class (building facades, roof, ground, vegetation, fields, and water). Performance of learned filtering is evaluated on several large SfM point clouds of cities. We find that results confirm our underlying assumption that discriminatively learning inlier-outlier distributions does improve precision over global heuristics by up to ≈ 12 percent points. Moreover, semantically informed filtering that models class-specific distributions further improves precision by up to ≈ 10 percent points, being able to remove very isolated building, roof, and water points while preserving inliers on building facades and vegetation. </jats:p> |
| doi_str_mv | 10.5194/isprs-annals-iv-2-263-2018 |
| facet_avail | Online |
| format | ElectronicArticle |
| format_de105 | Article, E-Article |
| format_de14 | Article, E-Article |
| format_de15 | Article, E-Article |
| format_de520 | Article, E-Article |
| format_de540 | Article, E-Article |
| format_dech1 | Article, E-Article |
| format_ded117 | Article, E-Article |
| format_degla1 | E-Article |
| format_del152 | Buch |
| format_del189 | Article, E-Article |
| format_dezi4 | Article |
| format_dezwi2 | Article, E-Article |
| format_finc | Article, E-Article |
| format_nrw | Article, E-Article |
| geogr_code | not assigned |
| geogr_code_person | not assigned |
| id | ai-49-aHR0cDovL2R4LmRvaS5vcmcvMTAuNTE5NC9pc3Bycy1hbm5hbHMtaXYtMi0yNjMtMjAxOA |
| imprint | Copernicus GmbH, 2018 |
| imprint_str_mv | Copernicus GmbH, 2018 |
| institution | DE-D275, DE-Bn3, DE-Brt1, DE-Zwi2, DE-D161, DE-Gla1, DE-Zi4, DE-15, DE-Pl11, DE-Rs1, DE-105, DE-14, DE-Ch1, DE-L229 |
| issn | 2194-9050 |
| issn_str_mv | 2194-9050 |
| language | English |
| last_indexed | 2024-03-01T17:05:24.224Z |
| match_str | stucker2018supervisedoutlierdetectioninlargescalemvspointcloudsfor3dcitymodelingapplications |
| mega_collection | Copernicus GmbH (CrossRef) |
| physical | 263-270 |
| publishDate | 2018 |
| publishDateSort | 2018 |
| publisher | Copernicus GmbH |
| record_format | ai |
| recordtype | ai |
| series | ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences |
| source_id | 49 |
| spelling | Stucker, C. Richard, A. Wegner, J. D. Schindler, K. 2194-9050 Copernicus GmbH General Medicine http://dx.doi.org/10.5194/isprs-annals-iv-2-263-2018 <jats:p>Abstract. We propose to use a discriminative classifier for outlier detection in large-scale point clouds of cities generated via multi-view stereo (MVS) from densely acquired images. What makes outlier removal hard are varying distributions of inliers and outliers across a scene. Heuristic outlier removal using a specific feature that encodes point distribution often delivers unsatisfying results. Although most outliers can be identified correctly (high recall), many inliers are erroneously removed (low precision), too. This aggravates object 3D reconstruction due to missing data. We thus propose to discriminatively learn class-specific distributions directly from the data to achieve high precision. We apply a standard Random Forest classifier that infers a binary label (inlier or outlier) for each 3D point in the raw, unfiltered point cloud and test two approaches for training. In the first, non-semantic approach, features are extracted without considering the semantic interpretation of the 3D points. The trained model approximates the average distribution of inliers and outliers across all semantic classes. Second, semantic interpretation is incorporated into the learning process, i.e. we train separate inlieroutlier classifiers per semantic class (building facades, roof, ground, vegetation, fields, and water). Performance of learned filtering is evaluated on several large SfM point clouds of cities. We find that results confirm our underlying assumption that discriminatively learning inlier-outlier distributions does improve precision over global heuristics by up to ≈ 12 percent points. Moreover, semantically informed filtering that models class-specific distributions further improves precision by up to ≈ 10 percent points, being able to remove very isolated building, roof, and water points while preserving inliers on building facades and vegetation. </jats:p> SUPERVISED OUTLIER DETECTION IN LARGE-SCALE MVS POINT CLOUDS FOR 3D CITY MODELING APPLICATIONS ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences |
| spellingShingle | Stucker, C., Richard, A., Wegner, J. D., Schindler, K., ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, SUPERVISED OUTLIER DETECTION IN LARGE-SCALE MVS POINT CLOUDS FOR 3D CITY MODELING APPLICATIONS, General Medicine |
| title | SUPERVISED OUTLIER DETECTION IN LARGE-SCALE MVS POINT CLOUDS FOR 3D CITY MODELING APPLICATIONS |
| title_full | SUPERVISED OUTLIER DETECTION IN LARGE-SCALE MVS POINT CLOUDS FOR 3D CITY MODELING APPLICATIONS |
| title_fullStr | SUPERVISED OUTLIER DETECTION IN LARGE-SCALE MVS POINT CLOUDS FOR 3D CITY MODELING APPLICATIONS |
| title_full_unstemmed | SUPERVISED OUTLIER DETECTION IN LARGE-SCALE MVS POINT CLOUDS FOR 3D CITY MODELING APPLICATIONS |
| title_short | SUPERVISED OUTLIER DETECTION IN LARGE-SCALE MVS POINT CLOUDS FOR 3D CITY MODELING APPLICATIONS |
| title_sort | supervised outlier detection in large-scale mvs point clouds for 3d city modeling applications |
| title_unstemmed | SUPERVISED OUTLIER DETECTION IN LARGE-SCALE MVS POINT CLOUDS FOR 3D CITY MODELING APPLICATIONS |
| topic | General Medicine |
| url | http://dx.doi.org/10.5194/isprs-annals-iv-2-263-2018 |