Stabilizing Videos By Removing Camera Shake Effect

Abstract

Home videos are usually suffering from jitter and high frequency unintended movements as most are taken with hand-held camcorders which makes necessary stabilize the footage. Given an un-stabilized video sequence, the objective of this project is to synthesize a new sequence as seen from a stabilized camera trajectory. Over the years there have been many different video stabilization techniques developed to solve this problem. However the removal of the unwanted video perturbations due to undesirable camera motions has proved difficult. The developed algorithm is based on a two-dimensional feature-based motion estimation technique. The method tracks a small set of features and estimates the movement of the camera between consecutive frames. An affine motion model is utilized to determine the parameters of translation and rotation between images. The determined affine transformation is then exploited to compensate for the abrupt temporal discontinuities of input image sequences. The algorithm is developed using OpenCV (Open Source Computer Vision) library with C++ interface. The algorithm is originally developed and tested on Windows platform. The algorithm also works under Linux environment. The project successfully generates a stable video sequence for a given unstable video sequence. The present solution to the problem removes unwanted affine transformation. Results show that the implementation is robust and can work in different scenarios. Moreover, it is possible to run it in a general purpose computer with good performance.