COMPUTATIONALLY EFFECTIVE NUMERICAL SIMULATION METHOD FOR DIFFRACTION-BLURRED IMAGES OF OBJECTS WITH PIECEWISE-LINEAR EDGE CONTOUR

Abstract

ubject of Research. The paper presents research of numerical image simulation for objects with known edge contour, considering optical diffraction blur and image sensor spatial sampling. Computationally effective solution is proposed under restriction of edge contour approximation by a set of straight lines.Method. Proposed method is based on the analytical numerical computation of the Fourier transform for the object bounding polygon. Calculus bandwidthis defined by the optical system diffraction limit and spatial frequency response of the object and background textures.Main Result. Proposed technical solution results in 2-3 order less computation time compared to a subpixel image simulation in spatial domain with subpixel sampling step value of 10-2 pixel.Computational complexity of the proposed approach is invariant with respect to a subpixel accuracy of simulated object geometry representation.Practical Relevance. Proposedapproach may be applied to a solution of inverse problems in the field of optical geometrical measurements and to a study of image processing algorithms with geometrical parameters representation error on the order of 10-1-10-4 pixel.

Keywords

• image simulation
• Fourier transform calculus
• subpixel edge location
• center of gravity method
• diffraction limit
• parallel computing