No-reference Image Sharpness Measure using Discrete Cosine Transform Statistics and Multivariate Adaptive Regression Splines for Robotic Applications
Kanjar De, Vedhanayagam Masilamani · Procedia Computer Science · 2018
Over the years the volume of digital data in the form of images and video have grown exponentially for robotic applications and assessing the quality of images has become an active area of research. One of the most common distortion in images is blur which can occur due to many sources like camera shake, defocus. Machine learning techniques have gained popularity in the area of image quality assessment in the past few years. In this paper, we propose a technique which uses Multivariate adaptive regression splines for predicting the quality of an image corrupted by image blur using Discrete Cosine Transforms statistics.