Video deraining in complex scene is a hot but challenging research topic. This paper proposes a novel video deraining network named STformer, which is integrating with spatial transformers and multiscale feature extraction. Specifically, the STformer architecture mainly comprises three primary components: a Local Feature Dynamic Extraction Network (LFDE) for preprocessing, a hierarchical encoder-decoder backbone with Spatial Transformer Blocks (STB) for feature extraction, and a Residual Mixture of Experts Feature Compensator (ResMEFC) for enhancing model performance and robustness. Especially, the proposed STB incorporates Channel-Wise Sparse Attention (CWSA) and Spatial Transformer Feedforward Network (STFN), and could focus on pertinent features for video deraining while minimizing noise interference. Extensive experiments on various benchmarks, including synthetic datasets like Rain200L/H and real-world datasets like SPA-Data and NTURain, demonstrate STformer’s superior performance to state-of-the-arts, particularly in terms of PSNR and SSIM.

This letter introduces an underwater image enhancement method to handle low contrast and color cast of underwater images. Firstly, with the help of hierarchical searching technique, we propose a novel backscattered light estimation method. And in this procedure, a novel scoring formula is considered into our method, which comprehensively considers multiple prior knowledge. Then, we generalize underwater dark channel prior (UDCP) approach to obtain more robust transmission estimation. In addition, we also develop a white balance method to further modify the appearance of the resultant image. Extensive experiments on real-world images demonstrate that the proposed method outperforms several previous image restoration or enhancement works.