| CPC G06V 10/98 (2022.01) [G02B 27/0172 (2013.01); G02B 30/33 (2020.01); G02F 1/133611 (2013.01); G06N 3/04 (2013.01); G06N 3/08 (2013.01); G06T 5/50 (2013.01); G06T 5/60 (2024.01); G06T 5/92 (2024.01); G06V 10/143 (2022.01); G02B 27/017 (2013.01); G02B 2027/0178 (2013.01); G06T 5/00 (2013.01); G06T 2207/10024 (2013.01); G06T 2207/20081 (2013.01); G06T 2207/20084 (2013.01); G06T 2207/20221 (2013.01)] | 21 Claims |

|
1. An image processing method comprising:
obtaining a plurality of input optical images and a plurality of output optical images for a near-eye display apparatus, the plurality of output optical images being output by a waveguide optical system provided in the near-eye display apparatus when the plurality of input optical images is input to the waveguide optical system;
training a neural network for modeling the waveguide optical system by using a training set comprising the obtained plurality of input optical images and the obtained plurality of output optical images; and
obtaining, when the training of the neural network is completed, a uniformity correction image to be input to the trained neural network such that the trained neural network then outputs a target uniformity image having a certain brightness uniformity,
wherein the trained neural network is obtained by modeling, into an image processing model, optical propagation in the waveguide optical system of the near-eye display apparatus for reproducing the plurality of input optical images as the plurality of output optical images.
|