Search

Research on Noise Suppression Algorithm for CMOS Image Sensors Based on Deep Learning

Author(s)

Junren Shao

Affiliation(s)

Guilin University of Electronic Technology, Guilin, Guangxi, China

Abstract

With the widespread application of CMOS image sensors, image noise has become an important factor affecting image quality, especially in low light or high ISO environments. As the key to improving image quality, noise suppression technology has become an important research direction in the field of image processing. Traditional noise suppression algorithms use filtering methods to smooth images. Although they can reduce noise to a certain extent, they often lead to the loss of image details and have limited processing capabilities for complex noise types. In recent years, noise suppression methods based on deep learning have developed rapidly. Through models such as convolutional neural networks (CNNs), generative adversarial networks (GANs), and autoencoders, deep learning can more accurately identify and remove noise while effectively retaining image details, showing superior denoising performance than traditional methods. Although deep learning methods have shown excellent results in noise suppression, they consume large computing resources and have slow training and inference speeds. In summary, the application of deep learning in noise suppression has made significant progress and is playing an increasingly important role in image sensor technology.

Keywords

CMOS Image Sensor; Noise Suppression; Deep Learning; Convolutional Neural Network; Generative Adversarial Network

References

[1] Cao, Yuxiang et al. “Image segmentation for night-vision surveillance camera based on deep learning.” Other Conferences (2022). [2] Guan, Jianjun et al. “Robust SAR Image Despeckling by Deep Learning From Near-Real Datasets.” IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing 17 (2024): 2963-2979. [3] Luo, Yifu et al. “MSR-Net: A Novel Noise Elimination Method for Real CMOS Image Sensor.” IEEE Access 12 (2024): 78714-78725. [4] Liang, Xujia et al. “Deep Learning Method on Target Echo Signal Recognition for Obscurant Penetrating Lidar Detection in Degraded Visual Environments.” Sensors (Basel, Switzerland) 20 (2020): n. pag. [5] Swin et al. “Signal Processing Based Pile-up Compensation for Gated Single-Photon Avalanche Diodes.” (2018). [6] Hendrickson, Ben et al. “Wavelet Analysis of RTS Noise in CMOS Image Sensors Irradiated With High-Energy Photons.” IEEE Transactions on Nuclear Science 67 (2020): 1732-1737. [7] Saito, Wataru et al. “A Low Noise and Linearity Improvement CMOS Image Sensor for Surveillance Camera with Skew-Relaxation Local Multiply Circuit and On-Chip Testable Ramp Generator.” 2021 IEEE Asian Solid-State Circuits Conference (A-SSCC) (2021): 1-3. [8] Zhou, Tao et al. “Effective Sparsity-Prior Image Denoising Algorithm for CMOS Image Sensor in Ultra-Low Light Imaging Applications.” 2020 China Semiconductor Technology International Conference (CSTIC) (2020): 1-3. [9] Fan, Qiaoyun et al. “Plume Noise Suppression Algorithm for Missile-Borne Star Sensor Based on Star Point Shape and Angular Distance between Stars.” Sensors (Basel, Switzerland) 19 (2019): n. pag. [10] Gao, Jing et al. “A Real Noise Elimination Method for CMOS Image Sensor Based on Three-Channel Convolution Neural Network.” IEEE Sensors Journal 20 (2020): 11549-11555.