A low complexity image sensing method using pseudo-random scan and recursive reconstruction with radial basis functions

A novel, low complexity image sensing method, reminiscent of compressive sensing is proposed. The “image sensing with pseudo-random scan” (ISPRS) method allows building efficient smart cameras and reconstruction units and differs from compressive sensing in both measuring and reconstruction stages. Instead of calculating projections of image samples, a counting automaton selects randomly K pixels from the sensing array. In the reconstruction stage, an iterative algorithm with linear complexity in K is employed. Reconstruction of missing samples is done using nonlinear radial basis functions centered on the spatial positions of measured pixels in a recursive process. Consequently, the complexities of both sensing device (camera) and reconstruction units are substantially reduced.