This article considers the problem of reconstructing a often obtained by using multiple identical image sensors shifted high-resolution image from multiple undersampled, shifted, degraded from each other by subpixel displacements [9,10]. The resulting frames with subpixel displacement errors. This leads to a formulation high-resolution image reconstruction problem using a set of curinvolving a periodically shift-variant system model. The maximum a rently available image sensors is interesting because it is closely posteriori (MAP) estimation scheme is used subject to the assumption related to the design of high-definition television (HDTV) image that the original high-resolution image is modeled by a stationary sensors. CCD image sensor arrays, consisting of a rectangular Markov-Gaussian random field. The resulting MAP formulation is exarray of sensing elements, produce discrete images whose sampressed as a complex linear matrix equation, where the characterizing pling rate and resolution are determined by the physical size of matrix involves the periodic block Toeplitz with Toeplitz block (BTTB) blur matrix and banded-BTTB inverse covariance matrix associated the sensing elements (undersampled images generated by the use with the original image. By approximating the periodic-BTTB and the of oversized sensing elements are further degraded by blur and, banded-BTTB matrices with, respectively, the periodic block circulant inevitably, noise as well). If multiple CCD image sensors are with circulant block (BCCB) and the banded-BCCB matrices, it is shifted from each other by exact subpixel values, the task of shown that the computation-intensive MAP formulation can be dehigh-resolution image reconstruction reduces to a single image composed into a set of smaller matrix equations by using the tworestoration problem with linear shift-invariant [(LSI) also linear dimensional discrete Fourier transform. Exact solutions are also conspace-invariant] blur operator [10]. However, as perfect subpixel sidered through the use of the preconditioned conjugate gradient displacements are practically impossible, blur functions in algorithm. Computer simulations are given to illustrate the procedure.
multisensor high-resolution image reconstruction could be peri-