Study on Scaling and Rotation Invariance of IWSVR

Inspire by human vision, imaging with space-variant resolution (IWSVR) has the characteristic of invariance to rotation and scaling, but the match between the center of the field of view (FOV) and target centroid is very strict. To quantify this invariance in IWSVR, a novel evaluation function is proposed in this paper. The proposed function when compared with two existing methods of scale-invariant feature transform (SIFT) and biaxial projection similarity analysis, shows reduced error within a five-pixel range of eccentricity. The results motivate the application of rotation and scale invariance property of IWSVR in different applications.