The human eye is very sensitive to structure information and is insensitive to distortion in areas with high brightness and high contrast areas. In order to take the human visual perception into consideration, a loss function based on the SSIM indicator would be used, which is an index that integrates human subjective perception.The following two groups of images show the comparison of images generated with SSIM loss function and without SSIM loss function along with its clear image and input rainy image for better comparison. As shown in the above groups of figures, Figure 30, the employment of the SSIM indicator in loss function does not show improvements in removing raindrops, and the de-raining ability of the network is reduced to some extent.The first group of images shows that the network added SSIM loss function reduced the de-raining performance to some extent. But from the “zoom in” version of images focusing on the blue handrail we can see that, the method added SSIM loss recovers better details for the handrail than the method without SSIM loss.