太原理工大学学报2025，Vol.56Issue(6)：1110-1117,8.DOI:10.16355/j.tyut.1007-9432.20220989
融合视觉注意力的粉尘图像深度估计
Depth Estimation of Dust Images with Visual Attention
摘要
Abstract
[Purposes]Unlike smog,dust distribution is often non-uniform and local,and there is a lack of an effective model description for light scattering and absorption,which seriously affects the depth estimation that relies on image features.To address this issue,a method for depth estimation of dust images incorporating visual attention is proposed.[Methods]In this method,the distribution model of dust in the image was established,and the visual attention network module was designed ac-cordingly to obtain the attention map of dust area,so as to guide the depth estimation network to strengthen the image depth feature extraction of dust area.In addition,in order to enhance the deep fea-ture extraction capability of dust images,a multi-scale feature extraction module was designed.The above modules were added to the framework of generative adversarial network and combined with the loss function designed for the depth estimation of dust images for realizing the depth estimation of a single dust image.[Results]The experimental results on the NYU Depth v2 dataset show that the mean relative error,logarithmic mean error,and root mean square error obtained by this method are 0.189,0.052,and 0.508,respectively,which are better than those of the current advanced algorithms for dust images.The comparison experiment on real data also proves that this method has improved generalization ability.关键词
深度估计/粉尘图像/注意力图Key words
depth estimation/dust image/attention map分类
信息技术与安全科学引用本文复制引用
宋曜宇,王园宇,王勤,张彤..融合视觉注意力的粉尘图像深度估计[J].太原理工大学学报,2025,56(6):1110-1117,8.基金项目
山西省自然科学基金(201801D121142) （201801D121142）
山西省回国留学人员科研资助项目 （）
