Texture edge smoothing and sharpening algorithm based on iterative nonlocal guided model
Image smoothing and sharpening are crucial operations in image processing, underpinning a wide array of applications across computer vision, medical imaging, and remote sensing. These processes are essential for delineating object details from noise, which is vital in fields such as graphics, compu...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | en |
| Published: |
Accent Social and Welfare Society
2025
|
| Online Access: | http://eprints.utem.edu.my/id/eprint/29243/2/017410209202511371.pdf http://eprints.utem.edu.my/id/eprint/29243/ https://accentsjournals.org/PaperDirectory/Journal/IJATEE/2025/2/4.pdf http://dx.doi.org/10.19101/IJATEE.2024.111101397 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Image smoothing and sharpening are crucial operations in image processing, underpinning a wide array of applications
across computer vision, medical imaging, and remote sensing. These processes are essential for delineating object details from noise, which is vital in fields such as graphics, computational photography, and computer vision. Despite their importance, achieving an ideal balance between smoothing and sharpening is challenging due to trade-offs and the presence of various types of noise and irregularities in real-life images. Traditional methods, such as Gaussian or median filtering (MF) for smoothing and Laplacian or unsharp masking for sharpening, often introduce artifacts or fail to
preserve crucial details. This work proposes a cutting-edge image filter that used iterative non-local guided model (inLG),
designed to be edge-aware and minimize halo artifacts. The primary objective is to enhance texture edge smoothing
performance while preserving essential details and sharpening critical features in digital images. The filter's effectiveness is demonstrated through applications in image enhancement, evaluated through quantitative and qualitative, confirming its capability. The experimental results demonstrate the algorithm's superior performance, achieving a mean squared error (MSE) of 0.276, a peak signal-to-noise ratio (PSNR) of 59.82 dB, and a structural similarity index (SSIM) of 0.999. These results surpass traditional methods, offering a balanced trade-off between edge preservation and noise reduction. |
|---|