SurroundNet: Towards effective low-light image enhancement

Although Convolution Neural Networks (CNNs) have made substantial progress in the low-light image enhancement task, one critical problem of CNNs is paradox model complexity and performance. This paper presents a novel SurroundNet that only involves less than 150K parameters (about 80–98 percent size reduction compared to SOTAs) achieves very competitive The proposed network comprises several Adaptive Retinex Blocks (ARBlock), which can be viewed as extension Single Scale feature space. core our ARBlock an efficient illumination estimation function called Surround Function (ASF). It regarded general form surround functions implemented by convolution layers. In addition, we also introduce Low-Exposure Denoiser (LED) smooth before enhancement. We evaluate method on two real-world datasets. Experimental results demonstrate superiority submitted both performance against State-of-the-Art methods. code available at https://github.com/ouc-ocean-group/SurroundNet.