Motion deblurring and super-resolution from an image sequence

Abstract

In many applications, like surveillance, image sequences are of poor quality. Motion blur in particular introduces significant image degradation. An interesting challenge is to merge these many images into one high-quality, estimated still. We propose a method to achieve this. Firstly, an object of interest is tracked through the sequence using region based matching. Secondly, degradation of images is modelled in terms of pixel sampling, defocus blur and motion blur. Motion blur direction and magnitude are estimated from tracked displacements. Finally, a highresolution deblurred image is reconstructed. The approach is illustrated with video sequences of moving people and blurred script.

A Bayesian approach to adaptive video super resolution

“…The lack of constraints is often addressed by spatial priors on the high-res image [22]. Hardie et al [8] jointly estimated the translational motion and the high-res image, while Bascle et al [4] also considered the motion blur using an affine motion model. But these motion models are too simple to reflect the nature of real-world sequences.…”

An overview of computational photography

“…Another way is to encode the high frequency information of the moving objects or the relative motion between the camera and the scene in some sense, and perform better restoration. The various approaches have been proposed, including: interleaving a camera array in chronological order [84,125], using a hybrid array [126]; coded photography (e.g., coded exposure [31,[127][128][129][130][131], coded sampling [132,133,21](see Figure 5(b)), coded sensor motion [134,135]); temporal super-resolution, which estimates the point spread function (blur kernel) from an image sequence [136][137][138] or a single photograph [139] before deconvolution, or performs temporal super-resolution from a set of low rate videos [140,141], or even introduces hardware attachments [142].…”

Blurring-invariant Riemannian metrics for comparing signals and images

“…A common solution has been to deblur the images using one of the many techniques available for deblurring [1,3,6,7,4,8] and then compare the deblurred images. The problem with this approach is that deblurring is a difficult operation, at least relative to blurring.…”