Prototype for Integration of colorization and super resolution model using machine learning

Colorization is the process of converting black and white images to color digitally. On the other hand, super resolution is the process of enhancing the quality of low-resolution pictures by converting them into high resolution, while maintaining the original details of the picture. The idea behind our implementation is to combine these two machine learning models. The combined model will take in old, blurry, hazed images as input and produce colorized, detailed and high-resolution image as an output. This research work has utilized various machine learning and computer vision techniques to accomplish this task. For colorization, classification based deep learning model was utilized. For super resolution, Post-up-sampling convolutional neural networks have been used.

I. Introduction

An image is representation of the feelings, words and emotions connected with the people in the photograph and the ones taking it, but, how to convey the feelings and emotions that are trapped inside those old black and white photos from our grandparents' era. The idea behind this project is to develop a fully automatic approach that will generate realistic high-resolution colorized images and videos, while taking low quality black and white images. Our aim is to produce a naturally appealing image that will convey what it was meant to do and carry on the message that was dying inside of that image. Our model gives a new life to old images and that is why we called it “RENAISSANCE”.