Image and video blurring is a common problem that can appear due to several conditions. Finding the blurred regions can have a great significance for characterizing the video content and analyzing its credibility, and also understanding the ability for improving and correcting it. The goal of this project is finding regions in the video which are blurred due to motion blur. Many methods were examined for detecting and characterizing the blur, in the spatial domain and in the frequency domain, while exploiting properties and assumptions on the video content. An algorithm was developed using Support Vector Machine (SVM), which uses the outputs of these various methods as a feature vector for the training process. The most contributing features for correct detection of blurred and non-blurred regions were found using many simulations, and a tool was written to display the results visually on the video itself. The results of the algorithm are good, and it is able to characterize blurred and non-blurred regions well however in a coarse way spatially. We discuss the algorithm and its results, and offer improvements for a few central building blocks as future work.