Application of multi-correlation-scale measurement matrices in ghost imaging via sparsity constraints

Sampling and reconstruction techniques are of special interest and importance in ghost imaging. Up to now, the transverse correlation scale of measurement matrices are usually constant. This paper explores a new possibility of constructing highly efficient measurement matrices with multi-correlation scales. Comparisons between the simulational and experimental results show that the multi-correlation-scale measurement matrices are highly efficient and accurate in sampling and image reconstruction and have a better antinoise ability than the existing constant-correlation-scale measurement matrices.