期刊
INTERNATIONAL JOURNAL OF MACHINE LEARNING AND CYBERNETICS
卷 14, 期 6, 页码 2169-2180出版社
SPRINGER HEIDELBERG
DOI: 10.1007/s13042-022-01753-x
关键词
Progressive image dehazing; Combined residual block; Adaptive feature fusion strategy; Fine haze layer
In this paper, a progressive dehazing network (PDN) is proposed to gradually remove haze by constructing preliminary and fine dehazing modules. Experimental results show that the proposed method outperforms some state-of-the-art dehazing methods in terms of visual comparison and objective evaluation.
Image dehazing is of great importance and has been widely studied, as haze severely affects many high-level computer vision tasks. In this paper, by considering the gradual dissipation process of haze, a progressive dehazing network (PDN) is proposed. The proposed approach realizes haze removal step by step by constructing two main modules: the preliminary and fine dehazing modules. In the preliminary dehazing module, a combined residual block is first constructed to extract and enhance features of different levels. Then, an adaptive feature fusion strategy is designed to integrate these features and output the initial dehazing result. Aiming at the residual haze in the initial results, a fine dehazing module is constructed by simulating the last period of the haze dissipation process to further extract a fine haze layer. The final dehazing result is obtained by removing the fine haze layer from the initial dehazing result. Experimental results indicate that the proposed method is superior to some state-of-the-art dehazing methods in terms of visual comparison and objective evaluation.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据