Computer Vision and Pattern Recognition, pp. 2736–2744 (2016)įu, X., Huang, J., Zeng, D., Huang, Y., Ding, X., Paisley, J.: Removing rain from single images via a deep detail network. S.: Rain streak removal using layer priors. In: International Conference on Computer Vision, pp. Luo, Y., Xu, Y., Ji, H.: Removing rain from a single image via discriminative sparse coding. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. T., Yang, W., Su, J., Liu, J.: Attentive generative adversarial network for raindrop removal from a single image. Density-aware single image de-raining using a multi-stream dense network. Lan, X., Ye, M., Zhang, S., et al.: Modality-correlation-aware sparse representation for RGB-infrared object tracking. Experimental results show that our proposed method outperforms the existing methods proposed for image deraining. In the second stage, image inpainting is combined with an attention mechanism through HSNet to recover the areas covered by raindrops in the input image.
In the first stage, the HSNet is used to extract raindrop features to make a residual between raindrop image samples, i.e., input data and output of HSNet to obtain an image without raindrops. This network is a fusion of dense network blocks and a U-Net network. To resolve the problem, we propose a method based on a hierarchical supervision network (HSNet). However, they are lacking in maintaining the balance between raindrop removal and image inpainting. These approaches are based on a pixel-wise regression process. Existing methods in removing raindrops from images have encountered a key challenge, i.e., removing raindrops of different sizes and shapes while recovering the lost details.
0 kommentar(er)
