US 12,423,802 B2
Defect discrimination apparatus for printed images and defect discrimination method
Mitsuhiro Yamashita, Matsumoto (JP); Takahiro Kamada, Matsumoto (JP); Naoki Hagihara, Shiojiri (JP); Takuya Ono, Shiojiri (JP); and Yuko Yamamoto, Shiojiri (JP)
Assigned to Seiko Epson Corporation, Tokyo (JP)
Filed by SEIKO EPSON CORPORATION, Tokyo (JP)
Filed on Feb. 28, 2023, as Appl. No. 18/176,299.
Claims priority of application No. 2022-030675 (JP), filed on Mar. 1, 2022.
Prior Publication US 2023/0281797 A1, Sep. 7, 2023
Int. Cl. G06T 7/00 (2017.01)
CPC G06T 7/001 (2013.01) [G06T 2207/20081 (2013.01)] 15 Claims
OG exemplary drawing
 
1. A defect discrimination apparatus for printed images, comprising:
a learning model that has undergone machine learning using a teacher image, the teacher image containing a defect that may occur during printing and being associated with a defect species, the learning model being configured to output similarity for each defect species;
a target image acquisition section that acquires an image of printed matter, which has been printed, and that prepares a target image to be an inspection target;
a discriminator that, with respect to the target image, uses the learning model to acquire similarity of a defect present in the target image to the defect species, and that discriminates the defect present in the target image as at least one known defect species; and
a learning section that, when updating the learning model based on a discrimination result by the discriminator, causes the learning model to undergo machine learning for a defect species that is different from the discriminated defect species or that is associated with an unknown defect.