Dual-functional fractal-fractional Sobel operator for efficient image enhancement and edge detection
Document Type
Article
Publication Date
4-1-2026
Abstract
Low-light conditions pose significant challenges for both image enhancement and edge detection, often leading to loss of detail and poor edge preservation. Existing methods address these problems separately, limiting their efficiency in real-world applications. To overcome this, we propose a novel Generalized Grünwald-Letnikov (GL) Dual-Functional Fractal-Fractional Sobel Operator (DF3SOp), integrating a dual-parameter approach for enhanced edge detection. The method extends the classical Sobel operator by incorporating fractional-order differentiation for enhancement and an adaptive fractal parameter for edge detection. Image enhancement is performed in the HSV color space, where the Value (V) channel undergoes Retinex-based decomposition into reflectance (R) and illuminance (L), with the fractional Sobel mask applied to L for detail enhancement. The final output is obtained via Stationary Discrete Wavelet Transform fusion. For edge detection, the adaptive fractal Sobel mask dynamically adjusts pixel-wise parameters to enhance structured edges. The proposed DF3SOp operates in three modes: enhancement, edge detection, or both, based on application needs. Extensive evaluations on diverse low-light image datasets demonstrate significant improvements in image clarity, texture preservation, and edge accuracy over classical and state of the art techniques, validated through SSIM, entropy, contrast, and edge density metrics.
Keywords
Adaptive edge detection, Fractal-fractional calculus, Fusion-based enhancement, Grünwald-Letnikov differentiation, Sobel operator
Publication Title
Pattern Recognition
ISSN
0031-3203
DOI
10.1016/j.patcog.2025.112592
Recommended Citation
Gowtham, K.; Babu, N. Ramesh; Harshavarthini, S.; and Raveendran, P., "Dual-functional fractal-fractional Sobel operator for efficient image enhancement and edge detection" (2026). Research Publications (2026 to 2030). 785.
https://knova.um.edu.my/research_publications_2026_2030/785
Funders
Arthritis National Research Foundation
Volume
172
Publisher
Elsevier Ltd