3D Scanning and Multiple Point Cloud Registration with Active View Complementation for Panoramically Imaging Large-Scale Plants
3D scanning is a surface reconstruction and a data processing technique which has already been widely used in reverse engineering, agricultural identification and other fields. However, most of the Commercial Off-The-Shelf (COTS) 3D scanning instruments are now developed for small and medium-sized targets, and there lacks an effective means for large-scale objects. In this paper, a multi-scale 3D scanning scheme with active-recognition is presented, which can effectively realize 3D information reconstruction of large-scale targets and recognition. First, data collected by 3D scanners based on structured light technology, which are placed at multiple angles, can be aligned together to generate a complete point cloud. Second, in order to achieve the panoramic view of the target, a camera with higher precision is mounted on the robotic arm for close-up shooting of the areas that are not easy to accurately capture. Finally, multi-scale and multi-resolution target scanning are achieved through refined scanning and detail feature recognition. This method combines large-scale structured light scanning with close-range image acquisition, which has potentials in fruit picking positioning, flaw detection of large workpieces, and three-dimensional human body modeling.
Keywords3D scanning Multi-scale Target recognition Point cloud registration
This work was supported by a grant from the National Natural Science Foundation of China (No. 51775333) and Scientific Research Program of Shanghai Science and Technology Commission (No. 18391901000).
- 3.Foglia, M.M., Reina, G.: Agricultural robot for radicchio harvesting. J. Field Robot. 23(6–7), 363–377 (2010)Google Scholar
- 4.Kondo, N., Ting, K.C.: Robotics for bioproduction systems. American Society of Agricultural Engineers (1998)Google Scholar
- 6.Zhengyou, Z.: Flexible camera calibration by viewing a plane from unknown orientations. In: Proceedings of the Seventh IEEE International Conference on Computer Vision, Kerkyra, Greece, vol. 1, pp. 666–673 (1999)Google Scholar
- 7.Sheng, Y., Kai, Z., Ka, Z.: Dissipative treatment of 3D laser scanning point cloud after multi-station stitching. Bull. Surveying Mapp. 3, 28–30 (2010)Google Scholar
- 8.Wu, Y., Li, P., Yang, Q.: Point cloud denoising algorithm based on neighborhood average method. J. Jiangxi Univ. Technol. 40(01), 10–15 (2019)Google Scholar
- 9.Zhou, W., Ma, X., Zhang, L.: Three dimensional point cloud splicing of tree canopy based on multi-source camera. Acta Optica Sinica 34(12), 185–192 (2014)Google Scholar
- 10.Rusinkiewicz, S.: Efficient variants of the ICP algorithm. In: Proceedings of the 3DIM (2001)Google Scholar
- 11.Chi, Y., Yu, X., Luo, Z.: 3D point cloud matching based on principal component analysis and iterative closest point algorithm. In: 2016 International Conference on Audio, Language and Image Processing (ICALIP), Shanghai, pp. 404–408 (2016)Google Scholar
- 14.Harris, G., Stephens, M.: A combined corner and edge detector. In: Proceedings 4th Alvey Vision Conference, pp. 147–151 (1988). https://doi.org/10.5244/c.2.23