Infrared Image Deblurring Based on Generative Adversarial Networks

Algorithm 1

Deblurring method proposed in this work.

	Input: clear image IS; blurred image IB
Output: deblurred image and synthesized blurred image discriminator judgment result;
(1)	for epoch = 1, …, 200 do
(2)	Sample real clear image IS and real blurred image IB from the training dataset
(3)	IS is sent to the generator GS2B to generate a blurred image
(4)	IB is sent to the generator GB2S to generate a sharp image
(5)	is sent to the generator GB2S to reconstruct the sharp image
(6)	is sent to the generator GS2B to reconstruct the blurred image
(7)	Update the discriminator DS and DB
(8)	Update the generator GS2B and GB2S
(9)	end for