Infrared star image denoising using regions with deep reinforcement learning

Mixed Poisson–Gaussian noise exists in the star images and is difficult to be effectively suppressed via maximum likelihood estimation (MLE) method due to its complicated likelihood function. In this article, the MLE method is incorporated with a state-of-the-art machine learning algorithm in order to achieve accurate restoration results. By applying the mixed Poisson–Gaussian likelihood function as the reward function of a reinforcement learning algorithm, an agent is able to form the restored image that achieves the maximum value of the complex likelihood function through the Markov Decision Process (MDP). In order to provide the appropriate parameter settings of the denoising model, the key hyperparameters of the model and their influences on denoising results are tested through simulated experiments. The model is then compared with two existing star image denoising methods so as to verify its performance. The experiment results indicate that this algorithm based on reinforcement learning is able to suppress the mixed Poisson–Gaussian noise in the star image more accurately than the traditional MLE method, as well as the method based on the deep convolutional neural network (DCNN).

Download Full-text

R3L: Connecting Deep Reinforcement Learning To Recurrent Neural Networks For Image Denoising Via Residual Recovery

10.1109/icip42928.2021.9506323 ◽

2021 ◽

Author(s):

Rongkai Zhang ◽

Jiang Zhu ◽

Zhiyuan Zha ◽

Justin Dauwels ◽

Bihan Wen

Keyword(s):

Neural Networks ◽

Reinforcement Learning ◽

Image Denoising ◽

Recurrent Neural Networks

Download Full-text

Noise suppression algorithm of short-wave infrared star image for daytime star sensor

Infrared Physics & Technology ◽

10.1016/j.infrared.2017.08.002 ◽

2017 ◽

Vol 85 ◽

pp. 382-394 ◽

Cited By ~ 10

Author(s):

Wenjie Wang ◽

Xinguo Wei ◽

Jian Li ◽

Gangyi Wang

Keyword(s):

Noise Suppression ◽

Short Wave ◽

Star Sensor ◽

Star Image ◽

Infrared Star ◽

Short Wave Infrared

Download Full-text

Pixel-Wise Deep Reinforcement Learning Approach for Ultrasound Image Denoising

10.1109/ius52206.2021.9593591 ◽

2021 ◽

Author(s):

Piotr Jarosik ◽

Marcin Lewandowski ◽

Ziemowit Klimonda ◽

Michal Byra

Keyword(s):

Reinforcement Learning ◽

Image Denoising ◽

Ultrasound Image ◽

Learning Approach

Download Full-text

Star Sensor Denoising Algorithm Based on Edge Protection

Sensors ◽

10.3390/s21165255 ◽

2021 ◽

Vol 21 (16) ◽

pp. 5255

Author(s):

Kaili Lu ◽

Enhai Liu ◽

Rujin Zhao ◽

Hui Zhang ◽

Hong Tian

Keyword(s):

Image Denoising ◽

Estimation Error ◽

Experimental Results ◽

Star Sensor ◽

Side Window ◽

Star Image ◽

Centroid Estimation ◽

Single Pixel

Single-pixel noise commonly appearing in a star sensor can cause an unexpected error in centroid extraction. To overcome this problem, this paper proposes a star image denoising algorithm, named Improved Gaussian Side Window Filtering (IGSWF). Firstly, the IGSWF algorithm uses four special triangular Gaussian subtemplates for edge protection. Secondly, it exploits a reconstruction function based on the characteristic of stars and noise. The proposed IGSWF algorithm was successfully verified through simulations and evaluated in a star sensor. The experimental results indicated that the IGSWF algorithm performed better in preserving the shape of stars and eliminating the single-pixel noise and the centroid estimation error (CEE) value after using the IGSWF algorithm was eight times smaller than the original value, six times smaller than that after traditional window filtering, and three times smaller than that after the side window filtering.

Download Full-text