Rain Streaks Removal from Rainy Images Using Convolutional Neural Networks

Abstract

Image processing is very important thing to improve the contrast and enhance the degraded images of real world. The rainy images are blurred and contain unclear structure of the objects for that apply deep learning based method to recover the rainy image objects. In this paper, remove the rainy streaks from rainy image and improve the contrast of the rainy image for that apply deep learning method of RainyNet is based on the Convolutional Neural Network. The RainyNet is made up of the encoder-decoder structure is constitute the detail rectify the network to find out the final clear structure. Here, DarkRain trained synthesised dataset is take over to testify the proposed method to produce good results than state-of-art methods. The proposed RainyNet is achieved better results than existing methods in terms of PSNR and SSIM.