Comparative Evaluation of EfficientNetB7 Independent Component Analysis and Adaptive Filtering for Image Enhancement

View at publisher Summarise with VIRAT
Authors: R. Anusuya, Rajalingam Arumuganainar, J. R. Arunkumar, V Vijaya Baskar
Cited by: 0
References: 12

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Image enhancement techniques are critical for applications such as medical imaging, surveillance, and remote sensing that require high-quality images. This study examines and contrasts three methods: EfficientNetB7-based deep learning, Independent Component Analysis (ICA), and Adaptive Filtering. It accomplishes this by analysing parameters such as Mean Squared Error (MSE), Peak Signal-to-Noise Ratio (PSNR), Root Mean Square Error (RMSE), Structural Similarity Index Measure (SSIM), Universal Image Quality Index (UIQI), and Visual Information Fidelity (VIF). The experimental results show that Adaptive Filtering provides balanced performance, implying that it preserves structure. This is demonstrated by a good SSIM of 0.6461 and low error metrics (MSE = 0.0031, RMSE = 0.0556). ICA demonstrates higher perceptual quality by having a competitive UIQI and strong fidelity (VIF=0.9181). In contrast, EfficientNetB7 performs poorly on standard image quality criteria because it is unable to maintain both structural and perceptual integrity. ICA often produces the most dependable results, although adaptive filtering achieves a decent balance between quality and speed.

Fields of Study & Subjects

Image Enhancement Techniques Image and Signal Denoising Methods Advanced Image Processing Techniques

Publication details

Journal / source
International Conference on Communication, Computing and Electronics Systems (ICCCES)
Pages
835–839
Publication date
21 January 2026
Language
EN