Super-resolution reconstruction of image sequences

Abstract

ÐIn an earlier work, we have introduced the problem of reconstructing a super-resolution image sequence from a given low resolution sequence. We proposed two iterative algorithms, the R-SD and the R-LMS, to generate the desired image sequence. These algorithms assume the knowledge of the blur, the down-sampling, the sequences motion, and the measurements noise characteristics, and apply a sequential reconstruction process. It has been shown that the computational complexity of these two algorithms makes both of them practically applicable. In this paper, we rederive these algorithms as approximations of the Kalman filter and then carry out a thorough analysis of their performance. For each algorithm, we calculate a bound on its deviation from the Kalman filter performance. We also show that the propagated information matrix within the R-SD algorithm remains sparse in time, thus ensuring the applicability of this algorithm. To support these analytical results we present some computer simulations on synthetic sequences, which also show the computational feasibility of these algorithms.

Advances and challenges in super‐resolution

“…We review the method introduced previously (Farsiu et al, 2004b), which offers an extremely efficient recursive algorithm for dynamic Super-Resolution. Although such a recursive solution for Super-Resolution has been addressed before (Elad and Feuer, 1999), we now show the speedups applicable for the case of translational motion and common space-invariant blur. This simplified model empowers us to use the two step algorithm that was described in Section IV.A for solving the dynamic case.…”

“…Note that in order to consider the varying reliability of measurements gathered at each location, the weighting can also be applied on a pixel-bypixel basis [(Farsiu et al, 2004b)]. We use the L 2 norm to follow Elad and Feuer's (1999) model (a robust data fusion term using L 1 norm minimization is part of our ongoing work).…”

“…An efficient and intuitive approach of acquiring the high-resolution image is using weighted least square optimization (Elad and Feuer, 1999): (13) where ␥ is a parameter between 0 and 1. The weighting ␥ places more emphasis on recent image data than on previous data.…”

Multi-channel Reconstruction of Video Sequences from Low-Resolution and Compressed Observations

“…Although work has been devoted to the problem of reconstruction of one high resolution image from a sequence of low resolution ones (see for instance [1][2][3][4][5] and [6] for a review), not much work has been reported on the problem of increasing the resolution of a whole image sequence simultaneously (see however [7][8][9]). …”

Hallucinating Faces: TensorPatch Super-Resolution and Coupled Residue Compensation

“…A number of superresolution techniques have been proposed in recent years [1][2] [3][5] [6] [7][9] [10]. They can be categorized into two representative approaches: the first is to produce a superresolution image from a set of low-resolution ones based on learning [5] [7][10]; another approach is to model the texture as a Markov Random Field (MRF) [3][6] [7]. These methods are mainly applied to generic images.…”