Image dehazing algorithm based on multi-level feature fusion and attention mechanism
An image dehazing network based on attention mechanism and multi-level feature fusion is proposed.The algorithm extracts features of different scales through residual dense network and self-calibration convolution network,and then uses dual attention units and pixel attention to fuse and reconstruct features.At the same time,a loss function combining mean square error loss,edge loss and robustness loss function is adopted,which can better retain the detail features.The experimental results show that this algorithm has made certain improvement in the peak signal to noise ratio and structural similarity indicators compared with other defogging algorithms,and the defogging image has made better performance in subjective vision.
万方数据
维普
