Wide Area Video Motion Blur Elimination

This Small Business Technology Transfer Phase I project will investigate the feasibility and effectiveness of fusing image analysis with Global Position System (GPS) and Inertial Measurement Unit (IMU) information to eliminate motion blur from hi-resolution wide area videos. The key innovations in the proposed effort are: 1) a novel non Linear Shift-Invariant (LSI) motion blur estimation algorithm that estimates the space variant motion blur field from Ą-channel of the input color image, and 2) an efficient unified framework to deblur hi-resolution wide area image. In the proposed framework, estimates from non-LSI blur estimation algorithm are used not only to refine blur parameters obtained using the GPS and IMU measurements but also to account for different blurring of independently moving targets. An efficient patch-by-patch solution is proposed to handle high resolution imagery that incorporates the structure of Bayer pattern for blur estimation and correction. In addition, enabling technologies will be developed to investigate and correct the impact of several factors, including lens distortion, rolling shutter, hilly terrain, parallax, etc. The Phase I effort will include: development of enabling algorithms, implementation of the framework, demonstration of proof of concept, theoretical computational complexity analysis and quantitative evaluation of the proposed technologies.