CONVERTING DEGRADED IMAGES TO HIGH RESOLUTION IMAGES USING CONVOLUTIONAL NEURAL NETWORK

During uncertain times like natural disaster, satellite imagery plays a key role in navigating the places to view the most affected places, monitor the forthcoming situation and get better insights from the images. To achieve this, the quality of the satellitery image should satisfy the standard image quality. Presently, a wide range of approaches are proposed to restore the good quality image from distorted image using different Deep learning techniques. To solve the problem with degraded images, different methods like Full resolution convolutional neural network, Deep resolution convolution neural network and Very deep resolution neural network and PSNR (Peak signal to noise ratio) are utilized. Among all, PSNR provides better colors. But PSNR cannot provide better quality of textures. To overcome this problem, a model has been proposed to restore the image quality by using SRCNN (Super Resolution Convolution Neural Network). Super resolution is one of the image restoration techniques, where it will split each layer in order to perform end-to-end mapping in each layer to obtain a high-quality image from a diminished image. This proposed system will restore the better texture for an image when the input is given as degraded image and it further yields the similar scope image as output with High tenacity.