Fast and robust real time digital video stabilization with smear removal using integral projection curve warping technique

Abstract

The paper describes the fractional i.e. subpixel motion compensation for videos having very small amount of disturbances. This enables accurate localization of targets and proper calibration of critical instruments having drift phenomena due to electronics hardware, specially for defence applications and consumer electronics having very less jitters. To remove motion smear due to destabilization, point spread function(PSF) is required, which is computationally very costly. Proposed algorithm calculates PSF with already known motion vectors with very less computation. The performance of the proposed algorithm is evaluated with a synthetic ground truth. A known amount of smooth random jitter and its proportional motion smear is incorporated in a stable video and its corresponding real time destabilized synthetic video is produced, which is then stabilized with our algorithm. The results show, that the quality of videos are improved by between 2 to 9 dB. The proposed technique is suitable for a variety of applications in real time, due to its very low computational cost. The applications range from critical defence applications to consumer electronics like handycams, digital cameras, cellphones etc. The stabilized videos are observed to have improved with good visual quality.