Analysis of Transmitting RainImages Based on Polar Codes (Encoder –Decoder) Under Additive White Gaussian Noise

Abstract

The most difficult in computer vision applications is removing white noise and rain in image. White noise and rain frequently coexist in real-world road photographs, but traditional imaging technologies are incapable of resolving this issue. However, we proposed a network based on framework with three step distinct Image decomposition using guided filters and white noise based on polar codes and rainfall removal network, since restoring an image based on a model of atmospheric scattering using expected transmit and received plans and predicted rain removed images technique. Investigational results establish that our trained framework outperforms better design approaches on synthetic and real-world way to test rain-images. We quantitatively evaluate our model using peak signal-to-noise indicators, demonstrating that our methods have the highest PSNR values. However, the design proposed to demonstrate that our method is applicable to real world vision applications based on polar codes (encoder and decoders).