Image stabilization for a camera on a moving platform

On Missing Data Treatment for Degraded Video and Film Archives: A Survey and a New Bayesian Approach

“…However, a simple and practical solution is to use fast determinstic motion estimation schemes to yield candidates for motion, for example block matching or optic flow schemes. The marginal conditional probability of these candidates are then evaluated using (14). This idea has not been fully explored in the literature but does yield much practical benefit.…”

“…2) Given the current estimates for and , the image is scanned on a block basis. Within each block, the solution for motion is generated by selecting eight candidate motion vectors from surrounding spatial blocks and choosing the one which minimizes the following "energy" [arising from (14)] (16) is then measured as the variance of the prediction error in the block after the motion is selected. Note that as the occlusion variables are taken into account (removing pixel pairs that are occluded), the normalizing term changes accordingly.…”

A Motion Model Based Video Stabilisation Algorithm

“…By fitting an affine motion model, the motion parameters are computed and the currently stabilized ith video frame is based on the previously stabilized frame or the original ith frame. The ambiguity between image motion caused by 3D rotation and that caused by 3D translation (per frame) is solved by analysing the direction of motion vectors [9,10] and their standard deviation.…”