Dehazing Via Reinforcement learing

Idea：We model the dehazing problem as a Markov Decision Process (MDP) with several existing simple traditional image processing operations and prior knowledge-based dehazing methods as actions

We use a deep reinforcement learning framework Deep Q-Network(DQN)

Agent Network

1. The role of Agent Network : Exteract Future.

2. The structure of Agent Network : The agent network consists of four fully connected layers,and the first three layers are followed by a ReLu layer.

3. The effect of Agent Network : It can effectively capture the inside cue ,and promote the seclection of correct action during the descion sequence.

4. The updata of Agent Network : Via the Q-learning process by Eq.(2)

   1. Q\*(s,a) = real time reward r + optimal action a' selected at the next moment in the Q*(s',a')
   

5. The loss fuction of Agent Network

   1. The loss fuction of agent network is defined as

      

   2. The gradient of loss fuction is defined as

      

Action space A

1. Add a $DC{P}^{\ast }=min(DCP(x),x)$ to improve the performance in the white areas of the image.
2. Define Actions of 5 to 10,which transform color channel’s value , and action 11 represents that theimage dehazing processing should be done .

reword function $R_t$

1. Adopt both PSNR and SSIM metrics as the measurement

   

2. Based on L2,PSNR and SSIM , the artical design the reward fuction by :

Farmework reinforcement learing

Classification of reinforcement learing :

1. value based -> DQN
2. poli based
3. actor critic