System and method for deep learning image super resolution
AssigneeSamsung Electronics Co., Ltd.
InventorsMostafa El-Khamy, Jungwon Lee, Haoyu Ren
Life of this patent
Estimating the term…
Abstract
In a method for super resolution imaging, the method includes: receiving, by a processor, a low resolution image; generating, by the processor, an intermediate high resolution image having an improved resolution compared to the low resolution image; generating, by the processor, a final high resolution image based on the intermediate high resolution image and the low resolution image; and transmitting, by the processor, the final high resolution image to a display device for display thereby.
Claims20
A method for super resolution imaging, the method comprising:
- receiving, by a processor, a low resolution image;
- generating, by the processor, a first intermediate high resolution image having an improved resolution compared to the low resolution image, wherein the first intermediate high resolution image is generated by a first super resolution network using the low resolution image;
- generating, by the processor, a second intermediate high resolution image having an improved resolution compared to the low resolution image, wherein the second intermediate high resolution image is generated by a second super resolution network using the low resolution image;
- merging, by the processor, the first intermediate high resolution image and the second intermediate high resolution image;
- and
- generating, by the processor, a final high resolution image based on the merging,
- wherein the generating of the final high resolution image includes performing a weighted sum across a convoluted output based on the first super resolution network and the second super resolution network.
A system for super resolution imaging, the system comprising:
- one or more processors;
- and
- memory coupled to the one or more processors and having instructions stored thereon that causes the one or more processors to:
- receive a low resolution image;
- generate a first intermediate high resolution image having an improved resolution compared to the low resolution image, wherein the first intermediate high resolution image is generated by a first super resolution network using the low resolution image;
- generate a second intermediate high resolution image having an improved resolution compared to the low resolution image, wherein the second intermediate high resolution image is generated by a second super resolution network using the low resolution image;
- merge the first intermediate high resolution image and the second intermediate high resolution image;
- and
- generate a final high resolution image based on the merging,
- wherein the generating of the final high resolution image includes performing a weighted sum across a convoluted output based on the first super resolution network and the second super resolution network.
Dependencies are inferred from the claim text.