Contents I. Introduction
Images record and display useful visual information or details even though the original image may be degraded due to some flaws in the imaging and capturing process such as motion blur, noise, etc. Digital image restoration is used in astronomical imaging, media, medical imaging, filmography, security and surveillance, forensic science, coding image and video files, restoration of uniformly blurred television pictures and so forth. Techniques ranging from Artificial Neural Network and Convolutional Neural Networks to K-nearest Neighbors find application in tackling issues such as segmentation, thresholding, filtering etc. Filtering is concerned with removal of undesirable features from images. Image filtering is done to enhance the quality of an image either by removing specific features or by highlighting other features of the image. Image blur or noise may occur for many reasons such as shutter speed is too slow, or bad camera holding technique or aperture is too wide or due to miss focus or motion blur or blur due to movement of the subject or shake due to internal vibrations or due to depth of the field, dirty lens or miss focusing. Whatever be the reason for many uses we need to remove these blurs from images. Removing noise from images significantly improves the accuracy in different tasks. If you want to perform object detection, removing the noise from the image can have better accuracy. Here we apply different smoothing and edge enhancement filtering methods
Mean, Median, Gaussian, Bilateral, Scharr, Sobel, Laplacian and Canny filters on an image and assess the quality of the image in both cases using an image quality assessment technique called BRISQUE and by calculating the PSNR ratio of images.
