Enhancing Image Resolution and Denoising Using Autoencoder

Nowadays, in real time, image processing is involved in various sectors like security, health care, banking, and face recognition. While capturing an image, there is more chance of noise engaged with multiple aspects of the surroundings. To improve the quality of the image and to get better classification results, we need to clean the picture, which is called pre-processing of the image. For the past 30 years, there is tremendous research happening on image processing by many researchers. Deep learning-based autoencoders are producing better results with minimum loss. Image denoising can be achieved with autoencoder architecture. The denoised image is taken as input to the next level to improve the resolution. In this paper, we have considered the popular dataset fashion mnist to denoising the image, which includes the noise. We used back-to-back autoencoders to perform both image denoising and resolution enhancement. In this approach, we can do the pre-processing stage once on the dataset for both image denoising and enhancement of image resolution. We have used binary cross-entropy as loss function to evaluate the performance of the model, and later, we have focussed on improving the resolution to the image. Denoising of an image followed by resolution enhancements in the same process minimizes the time and pre-processing steps separately.