Experimental Study on Selection of Aperture Pattern Based on Image Statistics

This paper presents optimal aperture pattern selection method, based on the statistical model of images and Kullback-Leiber (KL) divergence algorithm. The approach is verified by experimental analysis. Different aperture patterns have different sensitivities for depth information and different abilities to distinguish between depths, that is, it's helpful to distinguish and extract depth information. The KL divergence is computed between the blurry image distributions, which obtained by the statistical model of image, in the frequency domain at any two depths, and evaluated the comprehensive ability of the pattern to distinguish the depth information through experiment. Compare to different patterns by this evaluation, the relative best one between them can be find. The method proposed is applied to select a relative good one between several aperture patterns and its feasibility is confirmed by experimental results.