頁籤選單縮合
題名 | In-depth Analysis of Single Stereo Image Pair with SIFT-RANSAC Algorithm= |
---|---|
作者 | Deng, Qing-long; Ke, Ming-da; Wu, Pei-jung; Chen, Chien-yue; |
期刊 | Journal of Science and Innovation |
出版日期 | 20130700 |
卷期 | 3:3 2013.07[民102.07] |
頁次 | 頁87-90 |
分類號 | 312.49 |
語文 | eng |
關鍵詞 | Depth analysis; Single-lens 3D camera; SIFT-RANSAC; |
英文摘要 | In this study, we applied a micro prism type single-lens 3D camera technology. The micro prism sheet was set in front of the lens as a dual view image capture device. The stereo image pair was grabbed due to once capture by single-lens 3D camera. The distance of object from the CCD was 200 mm to 600 mm. The Scale-Invariant Feature Transform-RANdomSAmple Consensus (SIFT-RANSAC) method was used for obtaining and estimating the depth profile of the captured object, while the parameters of overlap of image pair range size and view field angle should be put into estimation. The depth error ratio was less than 18.21% and the minimum error was less than 4% where the image grabbed distance was less than 450 mm. When the object distance was increased, the depth error would be larger. The function of the single-lens 3D camera was similar to human eyes when the object was put on infinity; the disparity of object was disappeared. |
本系統之摘要資訊系依該期刊論文摘要之資訊為主。