A study on interlaced sampling with unknown offsets

In this paper, interlaced sampling is discussed, for which signals/images are sampled several times by an identical sampling device like a CCD camera with slightly displaced locations. Since offset parameters are unknown, the problem becomes challenging. A typical example of this formulation is super-resolution image reconstruction from multiple low resolution images. A well-defined solution has already been devised for the case where the number of unknown parameters is less than or equal to the number of measurements. However, this condition easily fails in practical situations. Hence, this paper proposes a signal reconstruction method that provides stable solutions even if the condition does not hold. The key factors introduced here are a statistical assumption for target signals and the minimization of a cost function. Simulation results show that the proposed method performs much better than the conventional method.