Robust edge spread function construction methods to counter poor sample spacing uniformity in the slanted-edge method

Abstract

The slanted-edge method describes an algorithm for measuring the spatial frequency response (SFR) of digital imaging systems. The method can be applied to edges oriented at nearly any angle, but there are some angles that cause simplistic implementations of the algorithm to fail, or produce inaccurate measurements. These angle-dependent phenomena are demonstrated to stem from a lack of uniformity in supersample spacing in the edge spread function (ESF). Two well-known slanted-edge implementation variants are adapted to minimize edge orientation dependent errors. These robust slanted-edge implementations are demonstrated yield accurate measurements, regardless of edge orientation angle or moderate image noise.