软件导刊2025，Vol.24Issue(5)：186-193,8.DOI:10.11907/rjdk.241195
基于注意力机制的多特征图像去雾算法
Multi Feature Image Dehazing Algorithm Based on Attention Mechanism
摘要
Abstract
In environments with significant changes in scene depth and uneven distribution of haze concentration,traditional deep learning dehazing algorithms often struggle to accurately evaluate image depth,resulting in poor dehazing performance for distant objects and loss of de-tails for nearby objects.To this end,a multi feature image dehazing algorithm based on attention mechanism is proposed,which uses multiple attention mechanisms to focus on different feature regions of the image,achieving the goal of removing deep haze and improving image details.The algorithm first introduces an efficient preprocessing module to extract the primary features of the image;Then,the multi feature extraction module extracts key feature information at different levels in the image by integrating multiple attention mechanisms;Finally,the image recon-struction module utilizes feature fusion gate technology to integrate multiple features to restore images with higher clarity.The experimental re-sults show that the proposed algorithm can effectively remove distant haze and restore clear images,with significant improvements in quantita-tive evaluation indicators such as structural similarity and peak signal-to-noise ratio.关键词
图像去雾/注意力机制/多特征提取/图像重建Key words
image dehazing/attention mechanisms/multi feature extraction/image reconstruction分类
计算机与自动化引用本文复制引用
王思洪..基于注意力机制的多特征图像去雾算法[J].软件导刊,2025,24(5):186-193,8.基金项目
湖南省自然科学基金重大项目(2021JC0009) （2021JC0009）
