Three-dimensional point cloud semantic segmentation method based on local and non-local feature aggregation
The invention discloses a three-dimensional point cloud semantic segmentation method based on local and non-local feature aggregation. The method aims to solve the problems that an existing feature learning process is sensitive to noise, only local information interaction of adjacent areas is consid...
Gespeichert in:
| Hauptverfasser: | , , , , , , , |
|---|---|
| Format: | Patent |
| Sprache: | chi ; eng |
| Schlagworte: | |
| Online-Zugang: | Volltext bestellen |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| container_end_page | |
|---|---|
| container_issue | |
| container_start_page | |
| container_title | |
| container_volume | |
| creator | JIA HAITAO CHANG LE SONG YIYUN LENG GENG LUO XIN WEI ZUQI XU WENBO REGAZA ADUGNA TESFAYE |
| description | The invention discloses a three-dimensional point cloud semantic segmentation method based on local and non-local feature aggregation. The method aims to solve the problems that an existing feature learning process is sensitive to noise, only local information interaction of adjacent areas is considered, then global context is obtained through a hierarchical structure, feature learning from bottom to top is generally caused, accordingly, extracted feature information is affected by outliers, and feature redundancy exists. The invention provides a weighted summation method based on local information and non-local feature information. According to the method, a 3D point cloud is used as input, and relative coordinates are used as local features in a point local unit; in the non-local feature extraction module of the points, sampling points are used as query points, the relevancy between the sampling points and the whole point cloud in the layer is calculated through an attention mechanism, then MLP is carried o |
| format | Patent |
| fullrecord | <record><control><sourceid>epo_EVB</sourceid><recordid>TN_cdi_epo_espacenet_CN118587428A</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>CN118587428A</sourcerecordid><originalsourceid>FETCH-epo_espacenet_CN118587428A3</originalsourceid><addsrcrecordid>eNqNizsOwjAQBdNQIOAOywFShI9IiyIQFVV6tNgvjoWzG8XO_bGAA1C9N9LMsni1_QSU1g-Q6FU40KheEpmgs6WIgSV5k4_LRuKUHRqQerX05AhLmYOa3LFYEpXySx04zROInZvgPt26WHQcIja_XRXb66VtbiVGfSCObCBIj-ZeVfWxPh129Xn_j_MGFUlB1w</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>patent</recordtype></control><display><type>patent</type><title>Three-dimensional point cloud semantic segmentation method based on local and non-local feature aggregation</title><source>esp@cenet</source><creator>JIA HAITAO ; CHANG LE ; SONG YIYUN ; LENG GENG ; LUO XIN ; WEI ZUQI ; XU WENBO ; REGAZA ADUGNA TESFAYE</creator><creatorcontrib>JIA HAITAO ; CHANG LE ; SONG YIYUN ; LENG GENG ; LUO XIN ; WEI ZUQI ; XU WENBO ; REGAZA ADUGNA TESFAYE</creatorcontrib><description>The invention discloses a three-dimensional point cloud semantic segmentation method based on local and non-local feature aggregation. The method aims to solve the problems that an existing feature learning process is sensitive to noise, only local information interaction of adjacent areas is considered, then global context is obtained through a hierarchical structure, feature learning from bottom to top is generally caused, accordingly, extracted feature information is affected by outliers, and feature redundancy exists. The invention provides a weighted summation method based on local information and non-local feature information. According to the method, a 3D point cloud is used as input, and relative coordinates are used as local features in a point local unit; in the non-local feature extraction module of the points, sampling points are used as query points, the relevancy between the sampling points and the whole point cloud in the layer is calculated through an attention mechanism, then MLP is carried o</description><language>chi ; eng</language><subject>CALCULATING ; COMPUTING ; COUNTING ; PHYSICS</subject><creationdate>2024</creationdate><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktohtml>$$Uhttps://worldwide.espacenet.com/publicationDetails/biblio?FT=D&date=20240903&DB=EPODOC&CC=CN&NR=118587428A$$EHTML$$P50$$Gepo$$Hfree_for_read</linktohtml><link.rule.ids>230,308,776,881,25543,76294</link.rule.ids><linktorsrc>$$Uhttps://worldwide.espacenet.com/publicationDetails/biblio?FT=D&date=20240903&DB=EPODOC&CC=CN&NR=118587428A$$EView_record_in_European_Patent_Office$$FView_record_in_$$GEuropean_Patent_Office$$Hfree_for_read</linktorsrc></links><search><creatorcontrib>JIA HAITAO</creatorcontrib><creatorcontrib>CHANG LE</creatorcontrib><creatorcontrib>SONG YIYUN</creatorcontrib><creatorcontrib>LENG GENG</creatorcontrib><creatorcontrib>LUO XIN</creatorcontrib><creatorcontrib>WEI ZUQI</creatorcontrib><creatorcontrib>XU WENBO</creatorcontrib><creatorcontrib>REGAZA ADUGNA TESFAYE</creatorcontrib><title>Three-dimensional point cloud semantic segmentation method based on local and non-local feature aggregation</title><description>The invention discloses a three-dimensional point cloud semantic segmentation method based on local and non-local feature aggregation. The method aims to solve the problems that an existing feature learning process is sensitive to noise, only local information interaction of adjacent areas is considered, then global context is obtained through a hierarchical structure, feature learning from bottom to top is generally caused, accordingly, extracted feature information is affected by outliers, and feature redundancy exists. The invention provides a weighted summation method based on local information and non-local feature information. According to the method, a 3D point cloud is used as input, and relative coordinates are used as local features in a point local unit; in the non-local feature extraction module of the points, sampling points are used as query points, the relevancy between the sampling points and the whole point cloud in the layer is calculated through an attention mechanism, then MLP is carried o</description><subject>CALCULATING</subject><subject>COMPUTING</subject><subject>COUNTING</subject><subject>PHYSICS</subject><fulltext>true</fulltext><rsrctype>patent</rsrctype><creationdate>2024</creationdate><recordtype>patent</recordtype><sourceid>EVB</sourceid><recordid>eNqNizsOwjAQBdNQIOAOywFShI9IiyIQFVV6tNgvjoWzG8XO_bGAA1C9N9LMsni1_QSU1g-Q6FU40KheEpmgs6WIgSV5k4_LRuKUHRqQerX05AhLmYOa3LFYEpXySx04zROInZvgPt26WHQcIja_XRXb66VtbiVGfSCObCBIj-ZeVfWxPh129Xn_j_MGFUlB1w</recordid><startdate>20240903</startdate><enddate>20240903</enddate><creator>JIA HAITAO</creator><creator>CHANG LE</creator><creator>SONG YIYUN</creator><creator>LENG GENG</creator><creator>LUO XIN</creator><creator>WEI ZUQI</creator><creator>XU WENBO</creator><creator>REGAZA ADUGNA TESFAYE</creator><scope>EVB</scope></search><sort><creationdate>20240903</creationdate><title>Three-dimensional point cloud semantic segmentation method based on local and non-local feature aggregation</title><author>JIA HAITAO ; CHANG LE ; SONG YIYUN ; LENG GENG ; LUO XIN ; WEI ZUQI ; XU WENBO ; REGAZA ADUGNA TESFAYE</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-epo_espacenet_CN118587428A3</frbrgroupid><rsrctype>patents</rsrctype><prefilter>patents</prefilter><language>chi ; eng</language><creationdate>2024</creationdate><topic>CALCULATING</topic><topic>COMPUTING</topic><topic>COUNTING</topic><topic>PHYSICS</topic><toplevel>online_resources</toplevel><creatorcontrib>JIA HAITAO</creatorcontrib><creatorcontrib>CHANG LE</creatorcontrib><creatorcontrib>SONG YIYUN</creatorcontrib><creatorcontrib>LENG GENG</creatorcontrib><creatorcontrib>LUO XIN</creatorcontrib><creatorcontrib>WEI ZUQI</creatorcontrib><creatorcontrib>XU WENBO</creatorcontrib><creatorcontrib>REGAZA ADUGNA TESFAYE</creatorcontrib><collection>esp@cenet</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>JIA HAITAO</au><au>CHANG LE</au><au>SONG YIYUN</au><au>LENG GENG</au><au>LUO XIN</au><au>WEI ZUQI</au><au>XU WENBO</au><au>REGAZA ADUGNA TESFAYE</au><format>patent</format><genre>patent</genre><ristype>GEN</ristype><title>Three-dimensional point cloud semantic segmentation method based on local and non-local feature aggregation</title><date>2024-09-03</date><risdate>2024</risdate><abstract>The invention discloses a three-dimensional point cloud semantic segmentation method based on local and non-local feature aggregation. The method aims to solve the problems that an existing feature learning process is sensitive to noise, only local information interaction of adjacent areas is considered, then global context is obtained through a hierarchical structure, feature learning from bottom to top is generally caused, accordingly, extracted feature information is affected by outliers, and feature redundancy exists. The invention provides a weighted summation method based on local information and non-local feature information. According to the method, a 3D point cloud is used as input, and relative coordinates are used as local features in a point local unit; in the non-local feature extraction module of the points, sampling points are used as query points, the relevancy between the sampling points and the whole point cloud in the layer is calculated through an attention mechanism, then MLP is carried o</abstract><oa>free_for_read</oa></addata></record> |
| fulltext | fulltext_linktorsrc |
| identifier | |
| ispartof | |
| issn | |
| language | chi ; eng |
| recordid | cdi_epo_espacenet_CN118587428A |
| source | esp@cenet |
| subjects | CALCULATING COMPUTING COUNTING PHYSICS |
| title | Three-dimensional point cloud semantic segmentation method based on local and non-local feature aggregation |
| url | https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-21T12%3A10%3A23IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-epo_EVB&rft_val_fmt=info:ofi/fmt:kev:mtx:patent&rft.genre=patent&rft.au=JIA%20HAITAO&rft.date=2024-09-03&rft_id=info:doi/&rft_dat=%3Cepo_EVB%3ECN118587428A%3C/epo_EVB%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true |