Resolution improvement using sampling diversity

Abstract

Sampling diversity occurs under certain circumstances during multiple-frame imaging of a common object. When an undersampled sensor takes images without repeating the locations of the samples on the object, sampling diversity will occur. The undersampling causes the local phenomenon of aliasing, and small local differences between frames can be seen upon close inspection. Small object features near the sampling limit can exhibit radical changes and may even vanish if at Nyquist. The multiple frames can be combined with a Projection Onto Convex Sets (POCS) approach [1], resulting in significant resolution improvement. Under the assumption of linear transformation of coordinates, we discuss an algorithm that computes the result in FFT-time using the Fast Fractional Fourier Transform [2].