US 12,154,261 B2
Image defect detection method, electronic device and readable storage medium
Chih-Te Lu, New Taipei (TW); Wan-Jhen Lee, New Taipei (TW); and Chin-Pin Kuo, New Taipei (TW)
Assigned to HON HAI PRECISION INDUSTRY CO., LTD., New Taipei (TW)
Filed by HON HAI PRECISION INDUSTRY CO., LTD., New Taipei (TW)
Filed on Aug. 26, 2022, as Appl. No. 17/896,663.
Claims priority of application No. 202210654056.3 (CN), filed on Jun. 9, 2022.
Prior Publication US 2023/0401691 A1, Dec. 14, 2023
Int. Cl. G06T 7/00 (2017.01); G06T 5/50 (2006.01); G06T 7/11 (2017.01); G06V 10/28 (2022.01); G06V 10/74 (2022.01)
CPC G06T 7/0008 (2013.01) [G06T 5/50 (2013.01); G06T 7/11 (2017.01); G06V 10/28 (2022.01); G06V 10/761 (2022.01); G06T 2207/20224 (2013.01); G06V 2201/07 (2022.01)] 20 Claims
OG exemplary drawing
 
1. An image defect detection method applied to an electronic device, the method comprising:
acquiring an image to be detected, a defective image, and a non-defective image;
dividing the image to be detected into a plurality of detecting areas, generating a detection accuracy value for each of the plurality of detecting areas based on the defective image and the non-defective image, and obtaining a plurality of detection accuracy values;
constructing a plurality of learners;
selecting an autoencoder for each of the plurality of detection accuracy values from the plurality of learners;
obtaining a multi-scale autoencoder model corresponding to each of the plurality of detecting areas by training the corresponding autoencoder based on the non-defective image;
dividing each of the plurality of detecting areas into a plurality of detecting blocks, and obtaining a plurality of reconstructed image blocks by inputting each of the plurality of detecting blocks into the corresponding multi-scale autoencoder model;
calculating a reconstruction error value between each of the plurality of reconstructed image blocks and the corresponding detecting block;
generating a detection result of a product contained in the image to be detected based on the reconstruction error value corresponding to each detecting block.