Informatics Publishing Limited

M. Velammal ¹, Thiyam Ibungomacha Singh ², Nilesh Madhukar Patil ³, Subharun Pal ⁴

Affiliations
1 Department of Computer science and Technology, Karpagam College of Engineering, IN
2 Department of Computer Science and Engineering, Manipur Institute of Technology, Manipur University Campus, IN
3 Department of Computer Engineering, Dwarkadas Jivanlal Sanghvi College of Engineering, IN
4 Department of Computer Science and Engineering, Indian Institute of Technology Jammu, IN

Abstract

This paper introduces a novel approach for multiframe image restoration using Generative Adversarial Networks (GANs). Traditional image restoration techniques often struggle with handling complex degradation patterns and noise in images. In contrast, GANs have demonstrated remarkable capability in generating realistic and high-quality images. The proposed method leverages the power of GANs to restore multiframe degraded images by training the generator to learn the underlying clean image from a set of degraded frames. The discriminator collaborates with the generator to ensure the fidelity of the restored output. Experimental results on various datasets show that the proposed multiframe image restoration approach achieves superior performance compared to state-of-the-art methods in terms of image quality and fidelity.

Keywords

Multiframe, Image Restoration, Generative Adversarial Networks (GANs), Degradation Patterns, Fidelity.

Full Text

     

Anand Karuppannan ¹, K. Subba Reddy ², Nilesh Madhukar Patil ³, Chandra Mouli Venkata Srinivas Akana ⁴

Affiliations
1 Department of Electronics and Communication Engineering, Gnanamani College of Technology, IN
2 Department of Computer Science and Engineering, Prakasam Engineering College, IN
3 Department of Computer Engineering, SVKM Dwarkadas J Sanghvi College of Engineering, IN
4 Bonam Venkata Chalamayya Engineering College, IN

Abstract

In this research, a novel model for multispectral image classification and analysis, leveraging Spectral-Spatial Deep DenseNet Learning is presented. This proposed framework combines spectral and spatial information to enhance the discriminative power of deep neural networks, enabling accurate classification of multispectral images. We conduct extensive experiments on benchmark datasets, demonstrating the superior performance of our method compared to existing approaches. Furthermore, we provide a comprehensive analysis of the learned features, shedding light on the interpretability and effectiveness of our model for multispectral image analysis tasks.

Keywords

Spectral-Spatial, Deep DenseNet, Multispectral Image, Classification

Full Text