RGB-D Images for Object Segmentation, Localization and Recognition in Indoor Scenes using Feature Descriptor and Hough Voting
- Title
- RGB-D Images for Object Segmentation, Localization and Recognition in Indoor Scenes using Feature Descriptor and Hough Voting
- Author
- 김기범
- Keywords
- RGB-D object segmentation; saliency map; RGBD object recognition; Hough Voting; point cloud; feature; extraction and matching
- Issue Date
- 2020-03
- Publisher
- IEEE
- Citation
- 2020 17th International Bhurban Conference on Applied Sciences and Technology (IBCAST), Page. 290-295
- Abstract
- With the development in technology, object localization and recognition systems in RGB and depth have become the essential part of vision systems. There are different ways of objects localization and recognition which segment the whole scene in parts. RGB-D and depth information is best way for the extraction of objects of interest. Researchers and scientists from all our world, continuously trying to improve the vision and detection property of computer systems for improving the people lives. Till now, the computers can sense using sensors and get involved in interesting conversations but lack the capability of understanding a real scene like humans. In our system, we allow the machines to detect and recognize the indoor RGB-d scenes using novel methodology. Inspired by the significance of human-computer interaction, we have presented a technique to recognize and localize the multiple objects present in RGB-D indoor scenes taken from RGB-D object dataset. We have proposed Saliency map based RGB-D object segmentation along with the multiple features and Hough voting to evaluate the performance of our proposed system with other proposed systems. The proposed system should be used in autonomous driving systems, security systems, violence detection, traffic monitoring, games, defense and sports scenes.
- URI
- https://ieeexplore.ieee.org/document/9044545?arnumber=9044545&SID=EBSCO:edseeehttps://repository.hanyang.ac.kr/handle/20.500.11754/163050
- ISBN
- 978-1-7281-4675-1; 978-1-7281-4676-8
- ISSN
- 2151-1411
- DOI
- 10.1109/IBCAST47879.2020.9044545
- Appears in Collections:
- ETC[S] > 연구정보
- Files in This Item:
There are no files associated with this item.
- Export
- RIS (EndNote)
- XLS (Excel)
- XML