NOISE REDUCTION IN DIGITAL IMAGES BASED ON ORIGINAL RAW FILES USING NEURAL NETWORKS

Alyona A. Zamyshlyaeva, Artem L. Kim

Abstract


This paper presents a noise reduction method for RAW photo images, focusing on preserving the original information and improving the processing quality. Digital image processing is important for surveillance and vision systems where quality and detail play a key role. The proposed method is based on a combination of UNet and HQSNet neural networks. HQSNet performs semi-square partitioning of the input data, emphasizing key regions and reducing the dimensionality of less significant ones. UNet, in turn, efficiently processes the prepared data, preserving high granularity and tone transitions. The method is tested on real images, including complex night portrait and starry sky scenes, demonstrating high performance on MSE metrics and expert evaluations. Comparison with traditional methods, such as median and Bilateral filters, showed the superiority of the new approach in both noise removal quality and image detail preservation. The advantages of the method include preservation of dynamic range and the possibility of deep post-processing. The results obtained confirm its effectiveness in digital processing problems, which makes the development promising for application in automatic image analysis and enhancement systems.

Keywords


noise reduction; neural networks; UNet; HQSNet; RAW photo images

Full Text:

PDF

References


Digital Image Processing. - URL: https://en.wikipedia.org/wiki/Digital_image_processing

Fraser B. textit{Real World Camera Raw With Adobe Photoshop CS2: Industrial-Strength Production Techniques}, Peachpit Pr, San Francisco, 2007.

Ronneberger O., Fischer P., Brox T. U-Net: Convolutional Networks for Biomedical Image Segmentation, MICCAI 2015. textit{Lecture Notes in Computer Science}, 2015, vol. 9351, pp. 1-4. - URL: https://arxiv.org/abs/1505.04597

Wang A., Dalca A., Sabuncu M. Neural Network-based Reconstruction in Compressed Sensing MRI Without Fully-sampled Training Data, pp. 4-6. - URL: https://arxiv.org/abs/2007.14979

Weiche H., Ziqian B., Junyu L. Deep Learning, Machine Learning - Digital Signal and Image Processing: From Theory to Application, pp. 79-90. - URL: https://arxiv.org/abs/2410.20304

Shrestha S. Image Denoising using New Adaptive Based Median Filters, pp. 1-3. - URL: https://arxiv.org/abs/2007.14979

Browne M. Convolutional Neural Networks for Image Processing: An Application in Robot Vision. Advances in Artificial Intelligence, 16th Australian Conference on Artificial Intelligence, Perth, Australia, 2003. pp. 27-31.

Goodfellow I. Generative Adversarial Nets. Montreal, Curran Associates, Inc., 2014.

Mildenhall B. NeRF in the Dark: High Dynamic Range View Synthesis from Noisy Raw Images, pp. 1-4. - URL: https://arxiv.org/abs/2111.13679

Thibaud E. Joint Demosaicking and Denoising by Fine-Tuning of Bursts of Raw Images, pp. 4-6. - URL: https://arxiv.org/abs/1905.05092

Xiangyu X. Towards Real Scene Super-Resolution with Raw Images, pp. 2-4. - URL: https://arxiv.org/abs/1905.12156

Downey A. Think DSP: Digital Signal Processing in Python. O'Reilly, Sebastopol, 2016.

Kumar T., Mileo A., Brennan R. Image Data Augmentation Approaches: A Comprehensive Survey and Future directions, pp. 1-5. - URL: https://arxiv.org/abs/2301.02830

Discrete cosine transform. - URL: https://en.wikipedia.org/wiki/Discretecosinetransform

Gholami A. A Survey of Quantization Methods for Efficient Neural Network Inference, pp. 1-3. - URL: https://arxiv.org/abs/2103.13630

Lanlan C., Yap-Peng T. Hybrid Color Filter Array Demosaicking for Effective Artifact Suppression. Journal of Electronic Imaging, 2009, vol. 15, no. 1, pp. 1-17. DOI: 10.1117/1.2183325


Refbacks

  • There are currently no refbacks.


 Save