US 12,467,738 B2
Method and system for high-precision localization of surface of object
Xia Yang, Guangzhou (CN); Junyou Qin, Guangzhou (CN); Guisong Guo, Guangzhou (CN); and Xiaohu Zhang, Guangzhou (CN)
Assigned to Sun Vat-sen University, Guangzhou (CN)
Appl. No. 18/032,342
Filed by Sun Yat-sen University, Guangzhou (CN)
PCT Filed Nov. 25, 2021, PCT No. PCT/CN2021/133252
§ 371(c)(1), (2) Date Apr. 18, 2023,
PCT Pub. No. WO2023/060717, PCT Pub. Date Apr. 20, 2023.
Claims priority of application No. 202111193896.6 (CN), filed on Oct. 13, 2021.
Prior Publication US 2024/0369346 A1, Nov. 7, 2024
Int. Cl. G01B 11/00 (2006.01); G01B 11/26 (2006.01); G06T 7/73 (2017.01)
CPC G01B 11/002 (2013.01) [G01B 11/26 (2013.01); G06T 7/73 (2017.01); G06T 2207/30204 (2013.01)] 8 Claims
OG exemplary drawing
 
1. A method for high-precision localization of a surface of an object, comprising:
making marks with a pre-set distribution density on the surface of the object to obtain a marked object, wherein making the marks further comprises:
engraving the marks on the surface through a mark making method using the marks of a pre-set pattern to obtain the marked object, wherein the marks of the pre-set pattern comprise a line segment, a point, a circle, a square, a cross, and a shape of combination thereof;
the mark making method comprises laser engraving, printing, and etching; and
the marks have an engraving pitch less than half of an image frame;
photographing the marked object and numbering the marks in an image to obtain an image with numbered marks, wherein photographing the marked object and numbering the marks in an image to obtain the image with numbered marks further comprises:
photographing and imaging the surface of the marked object to obtain a surface image;
processing the surface image through threshold segmentation to detect and obtain one or more target centers of all the marks in the surface image;
taking the one or more target centers as a reference point, taking the image of a pre-set pixel size as a mark feature image of the corresponding marks;
establishing a corresponding relationship between the mark feature image and the image with numbered marks using the principal component analysis method, reducing the mark feature image to a low dimension, and taking first 10 principal components of a principal component matrix as a one-dimensional vector representing the marks to determine a unique number; and
obtaining an image of the surface with numbered marks;
acquiring a relative position information about each of the marks on the surface of the object according to the image with the numbered marks and establishing a corresponding relationship between the image with numbered marks and the relative position information of each of the marks;
acquiring a real-time to-be-measured image and detecting marks in the real-time to-be-measured image to obtain to-be-measured marks; and
calculating a position of the surface of the object corresponding to the real-time to-be-measured image according to the position of the to-be-measured marks in the real-time to-be-measured image and the relative position information of the surface of the object corresponding to the marks.