Image Deblurring is a common restoration issue. However, existing deep learning approaches have generalization and interpretability issues. This research work provides a framework capable of regulated, confidence-based noise removal in this project to address these issues. The framework is built on merging two denoised images, both of which were generated from the same noisy input. One of the two is denoised using generic algorithms (for example, Gaussian), making few assumptions about the input images and generalizing across all cases. The other uses deep learning to denoise data and performs well on known datasets. Also, this research work presents a series of strategies for seamlessly fusing the two components in the frequency domain. Also, this research work presents a fusion technique that protects users from out-of-distribution inputs and estimates the confidence of a deep learning denoiser to allow users to interpret the result. Further, this research work will illustrate the efficacy of the suggested framework in various use cases through experiments.
|Original language||American English|
|Title of host publication||Proceedings of the 5th International Conference on I-SMAC (IoT in Social, Mobile, Analytics and Cloud), I-SMAC 2021|
|Publisher||Institute of Electrical and Electronics Engineers Inc.|
|Number of pages||7|
|State||Indexed - 2021|
|Event||5th International Conference on I-SMAC (IoT in Social, Mobile, Analytics and Cloud), I-SMAC 2021 - Palladam, India|
Duration: 11 Nov 2021 → 13 Nov 2021
|Name||Proceedings of the 5th International Conference on I-SMAC (IoT in Social, Mobile, Analytics and Cloud), I-SMAC 2021|
|Conference||5th International Conference on I-SMAC (IoT in Social, Mobile, Analytics and Cloud), I-SMAC 2021|
|Period||11/11/21 → 13/11/21|
Bibliographical notePublisher Copyright:
© 2021 IEEE.