US 12,260,595 B2
Multi-mode three-dimensional scanning method and system
Jun Zheng, Hangzhou (CN); Shangjian Chen, Hangzhou (CN); Jiangfeng Wang, Hangzhou (CN); and Lidan Zhang, Hangzhou (CN)
Assigned to SCANTECH (HANGZHOU) CO., LTD., Hangzhou (CN)
Appl. No. 17/757,695
Filed by SCANTECH (HANGZHOU) CO., LTD., Hangzhou (CN)
PCT Filed Dec. 17, 2020, PCT No. PCT/CN2020/137221
§ 371(c)(1), (2) Date Jun. 17, 2022,
PCT Pub. No. WO2021/121320, PCT Pub. Date Jun. 24, 2021.
Claims priority of application No. 201911301536.6 (CN), filed on Dec. 17, 2019.
Prior Publication US 2022/0383549 A1, Dec. 1, 2022
Int. Cl. G06T 7/521 (2017.01); G01B 11/00 (2006.01); G01B 11/25 (2006.01); G06T 7/80 (2017.01); G06T 15/04 (2011.01); G06T 7/30 (2017.01)
CPC G06T 7/85 (2017.01) [G01B 11/002 (2013.01); G01B 11/2513 (2013.01); G06T 7/521 (2017.01); G06T 15/04 (2013.01); G06T 7/30 (2017.01); G06T 2207/10028 (2013.01)] 9 Claims
OG exemplary drawing
 
1. A multi-mode three-dimensional scanning method, comprising:
obtaining intrinsic parameters and extrinsic parameters of a calibrated camera in different scanning modes, and upon switching between the different scanning modes, triggering a change of parameters of the camera to the intrinsic parameters and the extrinsic parameters in a corresponding scanning mode;
selecting and executing, according to a scanning requirement, the corresponding scanning mode that comprises:
a laser-based scanning mode that comprises: performing a laser projection, obtaining an image of 2D laser lines and markers, extracting information of the 2D laser lines and markers, reconstructing a three-dimensional laser line point cloud and markers by using a trigonometry or epipolar principle, and obtaining a conversion relationship between point clouds of a previous frame and of a subsequent frame according to matching information of existing markers, so as to obtain a surface point cloud of an object being scanned in a same coordinate system;
a speckle-based scanning mode that comprises: performing a speckle projection, obtaining a 2D speckle image and extracting information of the 2D speckle image, reconstructing a three-dimensional area array point cloud by using a template matching principle, and obtaining a conversion relationship between point clouds of a current frame and of a previous frame according to information of feature points of the existing point clouds, so as to obtain a surface point cloud of the object being scanned in a same coordinate system,
wherein selecting and executing, according to a scanning requirement, the corresponding scanning mode further comprises:
a transition scanning mode from the laser-based scanning mode to the speckle-based scanning mode that comprises: first using the laser-based scanning mode to obtain an image of 2D laser lines, reconstructing three-dimensional laser lines and markers, and obtaining a three-dimensional laser line point cloud according to a marker-based conversion relationship, then switching to the speckle-based scanning mode, reconstructing a three-dimensional area array point cloud based on a speckle image, matching features of the current three-dimensional area array point cloud according to features of the existing three-dimensional laser line point cloud to obtain a point cloud conversion relationship, and converting the current point cloud to the existing coordinate system according to the point cloud conversion relationship.