Sparse2Dense: From Direct Sparse Odometry to Dense 3-D Reconstruction
In this letter, we proposed a new deep learning based dense monocular simultaneous localization and mapping (SLAM) method. Compared to existing methods, the proposed framework constructs a dense three-dimensional (3-D) model via a sparse to dense mapping using learned surface normals. With single vi...
Gespeichert in:
| Veröffentlicht in: | IEEE robotics and automation letters 2019-04, Vol.4 (2), p.530-537 |
|---|---|
| Hauptverfasser: | , , |
| Format: | Artikel |
| Sprache: | eng |
| Schlagworte: | |
| Online-Zugang: | Volltext bestellen |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| container_end_page | 537 |
|---|---|
| container_issue | 2 |
| container_start_page | 530 |
| container_title | IEEE robotics and automation letters |
| container_volume | 4 |
| creator | Jiexiong Tang Folkesson, John Jensfelt, Patric |
| description | In this letter, we proposed a new deep learning based dense monocular simultaneous localization and mapping (SLAM) method. Compared to existing methods, the proposed framework constructs a dense three-dimensional (3-D) model via a sparse to dense mapping using learned surface normals. With single view learned depth estimation as prior for monocular visual odometry, we obtain both accurate positioning and high-quality depth reconstruction. The depth and normal are predicted by a single network trained in a tightly coupled manner. Experimental results show that our method significantly improves the performance of visual tracking and depth prediction in comparison to the state-of-the-art in deep monocular dense SLAM. |
| doi_str_mv | 10.1109/LRA.2019.2891433 |
| format | Article |
| fullrecord | <record><control><sourceid>proquest_RIE</sourceid><recordid>TN_cdi_ieee_primary_8605349</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><ieee_id>8605349</ieee_id><sourcerecordid>2298336455</sourcerecordid><originalsourceid>FETCH-LOGICAL-c329t-fd4b4cf22beb39d13a6f3a50ff22e4fc6b60705749e4e5bbc7ba0ad7e7997b563</originalsourceid><addsrcrecordid>eNpNkM1Lw0AQxRdRsNTeBS8LnlP3e7veStOqUCjUj-uSTSaaarN1N0H635uaUjzNMPN7j5mH0DUlY0qJuVuup2NGqBmziaGC8zM0YFzrhGulzv_1l2gU44YQQiXT3MgBmj_vshCBpVBHuMeL4Lc4rQLkDe43eFX4LTRhjxuP_yjMkxSvIfd1bEKbN5Wvr9BFmX1FGB3rEL0u5i-zx2S5eniaTZdJzplpkrIQTuQlYw4cNwXlmSp5JknZjUCUuXKKaCK1MCBAOpdrl5Gs0KCN0U4qPkRJ7xt_YNc6uwvVNgt767PKptXb1Prwbj-bD8sEN92HQ3Tb87vgv1uIjd34NtTdiZYxM-FcCSk7ivRUHnyMAcqTLyX2kK_t8rWHfO0x305y00sqADjhE0UkF4b_AuqPdYE</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2298336455</pqid></control><display><type>article</type><title>Sparse2Dense: From Direct Sparse Odometry to Dense 3-D Reconstruction</title><source>IEEE Electronic Library (IEL)</source><creator>Jiexiong Tang ; Folkesson, John ; Jensfelt, Patric</creator><creatorcontrib>Jiexiong Tang ; Folkesson, John ; Jensfelt, Patric</creatorcontrib><description>In this letter, we proposed a new deep learning based dense monocular simultaneous localization and mapping (SLAM) method. Compared to existing methods, the proposed framework constructs a dense three-dimensional (3-D) model via a sparse to dense mapping using learned surface normals. With single view learned depth estimation as prior for monocular visual odometry, we obtain both accurate positioning and high-quality depth reconstruction. The depth and normal are predicted by a single network trained in a tightly coupled manner. Experimental results show that our method significantly improves the performance of visual tracking and depth prediction in comparison to the state-of-the-art in deep monocular dense SLAM.</description><identifier>ISSN: 2377-3766</identifier><identifier>EISSN: 2377-3766</identifier><identifier>DOI: 10.1109/LRA.2019.2891433</identifier><identifier>CODEN: IRALC6</identifier><language>eng</language><publisher>Piscataway: IEEE</publisher><subject>Deep learning ; deep learning in robotics and automation ; Estimation ; Image reconstruction ; Machine learning ; Odometers ; Optical tracking ; Performance enhancement ; Predictions ; Reconstruction ; Simultaneous localization and mapping ; SLAM ; Three dimensional models ; Three-dimensional displays ; Training ; Visual-based navigation ; Visualization</subject><ispartof>IEEE robotics and automation letters, 2019-04, Vol.4 (2), p.530-537</ispartof><rights>Copyright The Institute of Electrical and Electronics Engineers, Inc. (IEEE) 2019</rights><lds50>peer_reviewed</lds50><woscitedreferencessubscribed>false</woscitedreferencessubscribed><citedby>FETCH-LOGICAL-c329t-fd4b4cf22beb39d13a6f3a50ff22e4fc6b60705749e4e5bbc7ba0ad7e7997b563</citedby><cites>FETCH-LOGICAL-c329t-fd4b4cf22beb39d13a6f3a50ff22e4fc6b60705749e4e5bbc7ba0ad7e7997b563</cites><orcidid>0000-0002-1170-7162 ; 0000-0002-7796-1438 ; 0000-0003-2482-3469</orcidid></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktohtml>$$Uhttps://ieeexplore.ieee.org/document/8605349$$EHTML$$P50$$Gieee$$H</linktohtml><link.rule.ids>230,314,776,780,792,881,27901,27902,54733</link.rule.ids><linktorsrc>$$Uhttps://ieeexplore.ieee.org/document/8605349$$EView_record_in_IEEE$$FView_record_in_$$GIEEE</linktorsrc><backlink>$$Uhttps://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-243927$$DView record from Swedish Publication Index$$Hfree_for_read</backlink></links><search><creatorcontrib>Jiexiong Tang</creatorcontrib><creatorcontrib>Folkesson, John</creatorcontrib><creatorcontrib>Jensfelt, Patric</creatorcontrib><title>Sparse2Dense: From Direct Sparse Odometry to Dense 3-D Reconstruction</title><title>IEEE robotics and automation letters</title><addtitle>LRA</addtitle><description>In this letter, we proposed a new deep learning based dense monocular simultaneous localization and mapping (SLAM) method. Compared to existing methods, the proposed framework constructs a dense three-dimensional (3-D) model via a sparse to dense mapping using learned surface normals. With single view learned depth estimation as prior for monocular visual odometry, we obtain both accurate positioning and high-quality depth reconstruction. The depth and normal are predicted by a single network trained in a tightly coupled manner. Experimental results show that our method significantly improves the performance of visual tracking and depth prediction in comparison to the state-of-the-art in deep monocular dense SLAM.</description><subject>Deep learning</subject><subject>deep learning in robotics and automation</subject><subject>Estimation</subject><subject>Image reconstruction</subject><subject>Machine learning</subject><subject>Odometers</subject><subject>Optical tracking</subject><subject>Performance enhancement</subject><subject>Predictions</subject><subject>Reconstruction</subject><subject>Simultaneous localization and mapping</subject><subject>SLAM</subject><subject>Three dimensional models</subject><subject>Three-dimensional displays</subject><subject>Training</subject><subject>Visual-based navigation</subject><subject>Visualization</subject><issn>2377-3766</issn><issn>2377-3766</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2019</creationdate><recordtype>article</recordtype><sourceid>RIE</sourceid><recordid>eNpNkM1Lw0AQxRdRsNTeBS8LnlP3e7veStOqUCjUj-uSTSaaarN1N0H635uaUjzNMPN7j5mH0DUlY0qJuVuup2NGqBmziaGC8zM0YFzrhGulzv_1l2gU44YQQiXT3MgBmj_vshCBpVBHuMeL4Lc4rQLkDe43eFX4LTRhjxuP_yjMkxSvIfd1bEKbN5Wvr9BFmX1FGB3rEL0u5i-zx2S5eniaTZdJzplpkrIQTuQlYw4cNwXlmSp5JknZjUCUuXKKaCK1MCBAOpdrl5Gs0KCN0U4qPkRJ7xt_YNc6uwvVNgt767PKptXb1Prwbj-bD8sEN92HQ3Tb87vgv1uIjd34NtTdiZYxM-FcCSk7ivRUHnyMAcqTLyX2kK_t8rWHfO0x305y00sqADjhE0UkF4b_AuqPdYE</recordid><startdate>20190401</startdate><enddate>20190401</enddate><creator>Jiexiong Tang</creator><creator>Folkesson, John</creator><creator>Jensfelt, Patric</creator><general>IEEE</general><general>The Institute of Electrical and Electronics Engineers, Inc. (IEEE)</general><scope>97E</scope><scope>RIA</scope><scope>RIE</scope><scope>AAYXX</scope><scope>CITATION</scope><scope>7SC</scope><scope>7SP</scope><scope>8FD</scope><scope>JQ2</scope><scope>L7M</scope><scope>L~C</scope><scope>L~D</scope><scope>ADTPV</scope><scope>AOWAS</scope><scope>D8V</scope><orcidid>https://orcid.org/0000-0002-1170-7162</orcidid><orcidid>https://orcid.org/0000-0002-7796-1438</orcidid><orcidid>https://orcid.org/0000-0003-2482-3469</orcidid></search><sort><creationdate>20190401</creationdate><title>Sparse2Dense: From Direct Sparse Odometry to Dense 3-D Reconstruction</title><author>Jiexiong Tang ; Folkesson, John ; Jensfelt, Patric</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c329t-fd4b4cf22beb39d13a6f3a50ff22e4fc6b60705749e4e5bbc7ba0ad7e7997b563</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2019</creationdate><topic>Deep learning</topic><topic>deep learning in robotics and automation</topic><topic>Estimation</topic><topic>Image reconstruction</topic><topic>Machine learning</topic><topic>Odometers</topic><topic>Optical tracking</topic><topic>Performance enhancement</topic><topic>Predictions</topic><topic>Reconstruction</topic><topic>Simultaneous localization and mapping</topic><topic>SLAM</topic><topic>Three dimensional models</topic><topic>Three-dimensional displays</topic><topic>Training</topic><topic>Visual-based navigation</topic><topic>Visualization</topic><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>Jiexiong Tang</creatorcontrib><creatorcontrib>Folkesson, John</creatorcontrib><creatorcontrib>Jensfelt, Patric</creatorcontrib><collection>IEEE All-Society Periodicals Package (ASPP) 2005-present</collection><collection>IEEE All-Society Periodicals Package (ASPP) 1998-Present</collection><collection>IEEE Electronic Library (IEL)</collection><collection>CrossRef</collection><collection>Computer and Information Systems Abstracts</collection><collection>Electronics & Communications Abstracts</collection><collection>Technology Research Database</collection><collection>ProQuest Computer Science Collection</collection><collection>Advanced Technologies Database with Aerospace</collection><collection>Computer and Information Systems Abstracts Academic</collection><collection>Computer and Information Systems Abstracts Professional</collection><collection>SwePub</collection><collection>SwePub Articles</collection><collection>SWEPUB Kungliga Tekniska Högskolan</collection><jtitle>IEEE robotics and automation letters</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Jiexiong Tang</au><au>Folkesson, John</au><au>Jensfelt, Patric</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Sparse2Dense: From Direct Sparse Odometry to Dense 3-D Reconstruction</atitle><jtitle>IEEE robotics and automation letters</jtitle><stitle>LRA</stitle><date>2019-04-01</date><risdate>2019</risdate><volume>4</volume><issue>2</issue><spage>530</spage><epage>537</epage><pages>530-537</pages><issn>2377-3766</issn><eissn>2377-3766</eissn><coden>IRALC6</coden><abstract>In this letter, we proposed a new deep learning based dense monocular simultaneous localization and mapping (SLAM) method. Compared to existing methods, the proposed framework constructs a dense three-dimensional (3-D) model via a sparse to dense mapping using learned surface normals. With single view learned depth estimation as prior for monocular visual odometry, we obtain both accurate positioning and high-quality depth reconstruction. The depth and normal are predicted by a single network trained in a tightly coupled manner. Experimental results show that our method significantly improves the performance of visual tracking and depth prediction in comparison to the state-of-the-art in deep monocular dense SLAM.</abstract><cop>Piscataway</cop><pub>IEEE</pub><doi>10.1109/LRA.2019.2891433</doi><tpages>8</tpages><orcidid>https://orcid.org/0000-0002-1170-7162</orcidid><orcidid>https://orcid.org/0000-0002-7796-1438</orcidid><orcidid>https://orcid.org/0000-0003-2482-3469</orcidid></addata></record> |
| fulltext | fulltext_linktorsrc |
| identifier | ISSN: 2377-3766 |
| ispartof | IEEE robotics and automation letters, 2019-04, Vol.4 (2), p.530-537 |
| issn | 2377-3766 2377-3766 |
| language | eng |
| recordid | cdi_ieee_primary_8605349 |
| source | IEEE Electronic Library (IEL) |
| subjects | Deep learning deep learning in robotics and automation Estimation Image reconstruction Machine learning Odometers Optical tracking Performance enhancement Predictions Reconstruction Simultaneous localization and mapping SLAM Three dimensional models Three-dimensional displays Training Visual-based navigation Visualization |
| title | Sparse2Dense: From Direct Sparse Odometry to Dense 3-D Reconstruction |
| url | https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-02-12T09%3A00%3A50IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest_RIE&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Sparse2Dense:%20From%20Direct%20Sparse%20Odometry%20to%20Dense%203-D%20Reconstruction&rft.jtitle=IEEE%20robotics%20and%20automation%20letters&rft.au=Jiexiong%20Tang&rft.date=2019-04-01&rft.volume=4&rft.issue=2&rft.spage=530&rft.epage=537&rft.pages=530-537&rft.issn=2377-3766&rft.eissn=2377-3766&rft.coden=IRALC6&rft_id=info:doi/10.1109/LRA.2019.2891433&rft_dat=%3Cproquest_RIE%3E2298336455%3C/proquest_RIE%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=2298336455&rft_id=info:pmid/&rft_ieee_id=8605349&rfr_iscdi=true |