Year: 2025
Type: conference-paper
Abstract: This paper addresses the challenges of low contrast, noise interference, and poor real-time performance in detecting industrial low-light images, we propose an enhanced algorithm that combines adaptive enhancement with the USM algorithm. Firstly, the approximate Gaussian filter denoising effect is realized by multiple mean filter iterations and the preprocessing speed is improved; then, the local mean square deviation and local mean value are used to improve the fixed parameters in Gamma enhancement to improve the adaptivity and enhancement efficiency; finally, the mean filter is used to replace the Gaussian fuzzy convolution processing in the USM algorithm, which improves the sharpening speed and enhances the image edges and contour clarity; The experimental results show that the improved algorithm achieves significant efficiency gains compared to the original algorithm, and when applied to industrial low-light images, it provides clearer edges and contours, laying a foundation for subsequent deep processing.
Authors: Yating Luo, Fei Wu
Institution: Wuhan Donghu University
Language: English
Topic: Image Enhancement Techniques
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
Funder: Wuhan Donghu University
OpenAlex ID: W4413812562