US 12,293,525 B2
Artificial intelligence enabled metrology
John Flanagan, Hillsboro, OR (US); Brad Larson, Portland, OR (US); and Thomas Miller, Portland, OR (US)
Assigned to FEI Company, Hillsboro, OR (US)
Filed by FEI Company, Hillsboro, OR (US)
Filed on Oct. 13, 2022, as Appl. No. 17/965,217.
Application 17/965,217 is a continuation of application No. 16/175,013, filed on Oct. 30, 2018, abandoned.
Prior Publication US 2023/0034667 A1, Feb. 2, 2023
Int. Cl. G06T 7/12 (2017.01); G06F 18/243 (2023.01); G06N 3/045 (2023.01); G06N 3/08 (2023.01); G06N 20/20 (2019.01); G06T 7/00 (2017.01); G06T 7/10 (2017.01); G06T 7/11 (2017.01); G06T 7/13 (2017.01); G06T 7/174 (2017.01); G06T 7/30 (2017.01); G06V 10/26 (2022.01); G06V 10/44 (2022.01); G06V 10/764 (2022.01); G06V 10/82 (2022.01)
CPC G06T 7/12 (2017.01) [G06F 18/243 (2023.01); G06N 3/045 (2023.01); G06N 3/08 (2013.01); G06N 20/20 (2019.01); G06T 7/0004 (2013.01); G06T 7/10 (2017.01); G06T 7/11 (2017.01); G06T 7/13 (2017.01); G06T 7/174 (2017.01); G06T 7/30 (2017.01); G06V 10/267 (2022.01); G06V 10/44 (2022.01); G06V 10/764 (2022.01); G06V 10/82 (2022.01); G06T 2207/10056 (2013.01); G06T 2207/10061 (2013.01); G06T 2207/20084 (2013.01); G06T 2207/30148 (2013.01)] 15 Claims
OG exemplary drawing
 
1. A method comprising:
segmenting a first image of structure into one or more classes to form an at least partially segmented image, the one or more classes including a boundary class, the boundary class identifying a boundary of the structure, wherein segmenting the first image into one or more classes includes classifying each pixel of the first image as belonging to one or more classes of a plurality of classes;
identifying boundary pixels in a second image based on a classification of the boundary pixels into the boundary class;
forming a virtual area around the identified boundary pixels;
performing analysis including applying an edge finding algorithm at the virtual area to identify an edge of the structure; and
performing metrology on the second image based on the edge of the structure,
wherein classifying each pixel of the first image as belonging to one or more classes of a plurality of classes comprises:
classifying pixels of an input image into a key points class by a first convolutional neural network; and
classifying pixels of the input image into a remainder of classes of the plurality of classes by a second convolutional neural network.