US 12,361,515 B2
Method and system of real-time super-resolution image processing
Noam Elron, Tel Aviv (IL); Alexander Itskovich, Haifa (IL); Shahar S Yuval, Haifa (IL); and Noam Levy, Karmiel (IL)
Assigned to Intel Corporation, Santa Clara, CA (US)
Filed by Intel Corporation, Santa Clara, CA (US)
Filed on Mar. 26, 2021, as Appl. No. 17/213,489.
Prior Publication US 2021/0233210 A1, Jul. 29, 2021
Int. Cl. G06T 3/4053 (2024.01); G06N 3/08 (2023.01); G06T 3/4046 (2024.01); G06T 5/50 (2006.01); G06T 5/70 (2024.01)
CPC G06T 3/4053 (2013.01) [G06N 3/08 (2013.01); G06T 3/4046 (2013.01); G06T 5/50 (2013.01); G06T 5/70 (2024.01)] 20 Claims
OG exemplary drawing
 
1. A method comprising:
obtaining low resolution (LR) images of a video sequence;
generating, with a first neural network, at least one of weights or biases based on statistics of the LR images, the at least one of the weights or the biases generated at a rate less than a frame rate of the video sequence;
providing the at least one of the weights or the biases to a second neural network different from the first neural network, the second neural network to implement a super-resolution (SR) neural network to perform high resolution (HR) image generation; and
generating, with the SR neural network, HR images based on the LR images at the frame rate of the video sequence.