Implementation of image reconstruction algorithm using compressive sensing in FPGA

Compressive Sensing (CS) is a technique that suggests the possibility of reconstruction of a signal vector using much smaller linear measurements than its dimension. Sparse signals are acquired in vectors using sensing matrices. If the signals are sparse enough the original signal can be reconstructed successfully. In CS applications while the signal can be acquired using basic methods, in reconstructing the signal using incomplete data sets high processing power and complex statistical computations are required. In this research OMP (Orthogonal Matching Pursuit) which is a faster and more hardware-implementable reconstruction algorithm among other methods is used. OMP algorithm is implemented on a Virtex-6 type FPGA (Field Programmable Gate Array). With various optimizations the designed system yielded at least thousand times faster results than CPU (Central Processing Unit) and GPU (Graphics Processing Unit) applications.