内容引导注意力融合的多尺度特征图像去雾算法  

Multi-scale Feature Image Defogging Algorithm Based on Content-guided Attention Fusion

作　　者：蒲亚亚 王彦博 苏勇东 徐忠承 PU Yaya;WANG Yanbo;SU Yongdong;XU Zhongcheng(School of Information Engineering,Chang’an University,Xi’an 710064,China)

机构地区：[1]长安大学信息工程学院,陕西西安710064

出　　处：《计算机与现代化》2025年第3期78-85,92,共9页Computer and Modernization

基　　金：陕西省自然科学基金面上项目(2022JM-056)。

摘　　要：针对当前去雾方法存在颜色失真、细节信息模糊等问题,本文基于编码器-解码器网络架构提出一种基于内容引导注意力融合的多尺度特征图像去雾算法。首先,采用多尺度特征提取模块进行编码,设计3个不同尺度并行的扩张卷积和SE注意力扩大感受野,提取不同尺度的特征,提高特征利用率。其次,在解码器中设计内容引导注意力融合模块动态赋予深层特征与浅层特征不同的权重,保留图像更多有效特征信息。最后,设计引入金字塔场景解析网络PSPNet提高全局信息获取的能力。实验结果表明,本文算法相比于其他几种算法在SOTS数据集上峰值信噪比和结构相似性分别平均提高了26.13%、6.39%,在真实含雾数据集上信息熵和平均梯度分别平均提高了3.27%、21.09%,改善了去雾不彻底和细节信息模糊问题。Aiming at the problems of color distortion and detail blur in current defogging methods,a multi-scale feature image defogging algorithm based on content-guided attention fusion is proposed with encoder-decoder network architecture.Firstly,multi-scale feature extraction module is used to encode,and three parallel expanded convolutions with different scales and squeeze and excitation attention are designed to enlarge the receptor field,extract features of different scales,and improve fea⁃ture utilization.Secondly,in the decoder,the content-guided attention fusion module is designed to dynamically improve differ⁃ent weights for the deep and the shallow features to retain more effective feature information.Finally,pyramid scene parsing net⁃work is introduced to improve the ability of global information acquisition.The experimental results show that compared with other algorithms,the proposed algorithm improves 26.13%and 6.39%on the peak signal-to-noise ratio and structural similarity of SOTS datasets,respectively.The entropy and average gradient of the real fog datasets are increased by 3.27%and 21.09%re⁃spectively.The proposed algorithm improves the problem of defog incompleteness and detail blur.

关 键 词：图像去雾 特征融合 注意力机制 多尺度特征 深度学习 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]