US 11,816,824 B2
Computer implemented process to enhance edge defect detection and other defects in ophthalmic lenses
Soon Wei Wong, Singapore (SG); and Kundapura Parameshwara Srinivas, Singapore (SG)
Assigned to EMAGE AI PTE LTD, Singapore (SG)
Filed by EMAGE AI PTE LTD, Singapore (SG)
Filed on Feb. 25, 2021, as Appl. No. 17/185,359.
Claims priority of application No. 10202001656V (SG), filed on Feb. 25, 2020.
Prior Publication US 2021/0264585 A1, Aug. 26, 2021
Int. Cl. G06T 7/00 (2017.01); G06N 3/08 (2023.01); G06T 5/00 (2006.01); G06T 7/13 (2017.01); G06T 7/11 (2017.01); G06N 3/042 (2023.01); G06N 3/045 (2023.01); G06V 10/25 (2022.01); G06V 10/764 (2022.01); G06V 10/26 (2022.01); G06V 10/94 (2022.01); G06V 10/82 (2022.01); G06V 10/44 (2022.01); G06V 10/32 (2022.01); G06V 20/66 (2022.01); G06V 10/34 (2022.01); G06V 20/52 (2022.01)
CPC G06T 7/0004 (2013.01) [G06N 3/042 (2023.01); G06N 3/045 (2023.01); G06N 3/08 (2013.01); G06T 5/002 (2013.01); G06T 7/11 (2017.01); G06T 7/13 (2017.01); G06V 10/25 (2022.01); G06V 10/26 (2022.01); G06V 10/32 (2022.01); G06V 10/34 (2022.01); G06V 10/454 (2022.01); G06V 10/764 (2022.01); G06V 10/82 (2022.01); G06V 10/94 (2022.01); G06V 20/52 (2022.01); G06V 20/66 (2022.01); G06T 2207/20021 (2013.01); G06T 2207/20081 (2013.01); G06T 2207/20084 (2013.01); G06T 2207/30108 (2013.01)] 10 Claims
OG exemplary drawing
 
1. A computer-implemented method for creating an object detection model using neural networks to increase inspection efficiency comprising at least one high performance processor, fast access memory and several parallel graphic processing units, the method comprising:
an uncompressed high resolution image of the object;
preprocessing the image by applying smoothing algorithms to enhance defect information without compressing the pixel data;
segmenting the pre-processed image into segments and performing dimension normalisation by reconstructing the segments into square areas, suited for input to the graphic processing unit;
applying deep learning algorithms to optimised images comprising the reconstructed segments to extract feature information within each image to aid in machine learning and training;
classifying the extracted and calibrated feature information under several categories through application of neural networks;
re-classifying the categories of extracted feature information from new images generated through application of generative adversarial networks to the optimised and uncompressed images;
creating an extensive database of extracted features detected within the optimized images for micro defect inspection in uncompressed images; and
applying the neural networks in combination with the database to inspect new devices at high speed and efficiency.