Residual Dense Network with an Attention Mechanism for Image Dehazing

In response to the common problems of incomplete dehazing and color distortion in existing image dehazing algorithms, this paper proposes an image dehazing algorithm based on attention mechanism and residual dense network. Firstly, design a shallow feature extraction module that uses multi-scale convolutional kernels to extract texture information of different scenes. Secondly, based on the encoding decoding model, a network structure is constructed using residual dense blocks and global residual connections, and combined with a dual channel attention mechanism module for feature fusion and extraction. The experimental results show that the proposed algorithm achieves higher peak signal-to-noise ratio and structural similarity compared to mainstream dehazing algorithms, and has good dehazing performance.