Adaptive Hybrid Wavelet Framework for Efficient Multi-Noise Image Denoising and Frequency-Aware Reconstruction

Authors

  • Arpit Urmaliya
  • J.k maitra

Keywords:

Adaptive Image Denoising, Wavelet Transform, Transformer Attention Enhancement, Frequency-Aware Reconstruction and Hybrid Image Processing Framework

Abstract

The image denoising is crucial to improve the quality of images in medical imaging, remote sensing, surveillance, and digital photography applications. The main aim of this research is to propose an adaptive hybrid wavelet-based image denoising framework which can effectively remove the Gaussian noise, Salt & Pepper noise and Speckle noise and preserve structural and texture information in the denoised image. The proposed methodology combines Discrete Wavelet Transform (DWT), Stationary Wavelet Transform (SWT), Wavelet Packet Transform (WPT), Dual Tree Complex Wavelet Transform (DTCWT) with adaptive threshold optimization, transformer-inspired attention enhancement and frequency-aware reconstruction techniques. The entire system is realized in MATLAB including automatic loading and processing of datasets, faults addition and adaptive selection of the optimum wavelet and the application of weighted hybrid fusion techniques. All the experimental analysis is done using PSNR, SSIM, RMSE, MAE, Entropy and Edge Preservation Index (EPI) metrics. It achieved a maximum PSNR of 22.6379 dB, an SSIM of 0.7508 and a minimum RMSE of 0.0738 under Gaussian noise conditions. The results with Speckle noise showed that the proposed method achieved the maximum PSNR value of 25.0669 dB with the maximum SSIM of 0.6083, and the DTCWT method achieved the maximum PSNR value of 25.8132 dB. The experimental results show that the proposed hybrid framework is able to achieve better denoising performance, preserve edges, and improve the reconstruction quality than traditional wavelet-based denoising methods.

References

L. Quinones and R. Tibi, “Denoising Seismic Waveforms Using a WaveletTransform-Based Machine-Learning Method,” Bull. Seismol. Soc. Am., vol. 114, no. 4, pp. 1777–1788, 2024, doi: 10.1785/0120230304.

H. Q. Birdawod, A. M. Khudhur, D. H. Kadir, and D. M. Saleh, “A Wavelet Shrinkage Mixed with a Single-level 2D Discrete Wavelet Transform for Image Denoising,” Kurdistan J. Appl. Res., vol. 9, no. 2, pp. 1–12, 2024, doi: 10.24017/science.2024.2.1.

C. Tamilselvi, M. Yeasin, R. K. Paul, and A. K. Paul, “Can Denoising Enhance Prediction Accuracy of Learning Models? A Case of Wavelet Decomposition Approach,” Forecasting, vol. 6, no. 1, pp. 81–99, 2024, doi: 10.3390/forecast6010005.

L. Wang, Z. Yang, Y. F. Pu, H. Yin, and X. Ren, “An Efficient Multi-Scale Wavelet Approach for Dehazing and Denoising Ultrasound Images Using Fractional-Order Filtering,” Fractal Fract., vol. 8, no. 9, pp. 1–16, 2024, doi: 10.3390/fractalfract8090549.

R. T. Sadia, J. Chen, and J. Zhang, “CT image denoising methods for image quality improvement and radiation dose reduction,” J. Appl. Clin. Med. Phys., vol. 25, no. 2, pp. 1–17, 2024, doi: 10.1002/acm2.14270.

Downloads

How to Cite

Arpit Urmaliya, & J.k maitra. (2025). Adaptive Hybrid Wavelet Framework for Efficient Multi-Noise Image Denoising and Frequency-Aware Reconstruction. International Journal of Engineering Science & Humanities, 15(4), 1173–1199. Retrieved from https://www.ijesh.com/j/article/view/1300