Data-based 3D shape reconstruction using light field constructed by multiple projectors

• Item type: SIG Technical Reports(1)
• 公開日: 2017-11-01
• タイトル: Data-based 3D shape reconstruction using light field constructed by multiple projectors
• タイトル / 言語: en
• タイトル: Data-based 3D shape reconstruction using light field constructed by multiple projectors
• 言語: eng
• キーワード / 主題Scheme: Other
• キーワード / 主題: 一般セッション5
• 資源タイプ / 資源タイプ識別子: http://purl.org/coar/resource_type/c_18gh
• 資源タイプ: technical report
• 著者所属: Kagoshima University
• 著者所属: Kagoshima University
• 著者所属: Hiroshima City University
• 著者所属: Hiroshima City University
• 著者所属: Kagoshima University
• 著者所属(英): en
• 著者所属(英): Kagoshima University
• 著者所属(英): en
• 著者所属(英): Kagoshima University
• 著者所属(英): en
• 著者所属(英): Hiroshima City University
• 著者所属(英): en
• 著者所属(英): Hiroshima City University
• 著者所属(英): en
• 著者所属(英): Kagoshima University
• 著者名: Yuki, Shiba; Satoshi, Ono; Ryo, Furukawa; Shinsaku, Hiura; Hiroshi, Kawasaki
• 著者名(英): Yuki, Shiba; Satoshi, Ono; Ryo, Furukawa; Shinsaku, Hiura; Hiroshi, Kawasaki
• 論文抄録 / 内容記述タイプ: Other
• 論文抄録 / 内容記述: Combination of a pattern projector and a camera is widely used for 3D measurement. From the pattern projection, various kinds of depth queues are extracted from the captured image such as disparities for active stereo, projector defocus for depth from defocus, or intensity variations for photometric stereo. To increase the depth queue information, while widening the working space and reducing the occlusions, increasing the number of projectors can be a promising solution. However, multiple projectors form a complicated light field where the captured image is difficult to process analytically. In this paper, we use the configuration of multiple projectors and a camera. To process the complicated light field, matching-based analysis is applied. In the preprocess of the 3D measurement, virtual sample images of planar board with various depths are generated with CG techniques. Then, their image features of small patches are extracted with PCA. In the 3D measurement step, the same image features of patches of the captured image are extracted and compared with the sample images. By utilizing the dimensional reduction of PCA and ANN search algorithm, the matching is processed effectively. Since our approach is sampling based, the proposed technique can handle arbitrary patterns for projection.
• 論文抄録(英) / 内容記述タイプ: Other
• 論文抄録(英) / 内容記述: Combination of a pattern projector and a camera is widely used for 3D measurement. From the pattern projection, various kinds of depth queues are extracted from the captured image such as disparities for active stereo, projector defocus for depth from defocus, or intensity variations for photometric stereo. To increase the depth queue information, while widening the working space and reducing the occlusions, increasing the number of projectors can be a promising solution. However, multiple projectors form a complicated light field where the captured image is difficult to process analytically. In this paper, we use the configuration of multiple projectors and a camera. To process the complicated light field, matching-based analysis is applied. In the preprocess of the 3D measurement, virtual sample images of planar board with various depths are generated with CG techniques. Then, their image features of small patches are extracted with PCA. In the 3D measurement step, the same image features of patches of the captured image are extracted and compared with the sample images. By utilizing the dimensional reduction of PCA and ANN search algorithm, the matching is processed effectively. Since our approach is sampling based, the proposed technique can handle arbitrary patterns for projection.
• 書誌レコードID / 収録物識別子タイプ: NCID
• 書誌レコードID / 収録物識別子: AA11131797
• 書誌情報: 研究報告コンピュータビジョンとイメージメディア（CVIM）; 巻 2017-CVIM-209, 号 33, p. 1-7, 発行日 2017-11-01
• ISSN / 収録物識別子タイプ: ISSN
• ISSN / 収録物識別子: 2188-8701
• Notice: SIG Technical Reports are nonrefereed and hence may later appear in any journals, conferences, symposia, etc.
• 出版者 / 言語: ja
• 出版者: 情報処理学会