Study of an iterative super-resolution algorithm and its feasibility in high-resolution animal imaging with low-resolution SPECT cameras

In this work, we describe the principle of a general- purpose iterative algorithm for super-resolution and present the ongoing study of its application in the improvement of the resolution of SPECT images. The iterative algorithm uses a set of images acquired with low-resolution and at slightly different positions to form a single high-resolution image. The algorithm was tested by simulating the acquisition, for a given test image with super-resolution, of a set of N² planar images obtained with a low-resolution camera located at N² positions forming a grid with a distance between nodes N times smaller than the spatial resolution of the camera. In the absence of noise, the algorithm converges to the exact solution, recovering the original super- resolution image. Tests performed with different levels of Poisson noise added to the acquired low-resolution images indicate that the algorithm converges to an image that is very similar to the original super-resolution image. These results suggest that it is possible to acquire very-high spatial resolution images even with very low-cost, low-resolution cameras, which could have applications for instance in animal imaging, where high resolution images are essential.