Evaluation of image processing algorithms

Abstract

One exemplary aspect of this invention pertains to a method to evaluate an image processing algorithm. The method includes varying a parameter of a model of an imaging system and, for each variation of the parameter, calculating with a data processor a corresponding image of a sample; applying an image processing algorithm to the calculated corresponding images of the sample; and determining an ability of the image processing algorithm to detect the variation in the parameter.

Description

technical field [0001] Illustrative embodiments of the present invention generally relate to a method of systematically and objectively evaluating and improving the effectiveness of image processing algorithms, thereby eliminating uncertainty and variability introduced by human judgment. Background technique [0002] Image processing algorithms typically compute a figure of merit for a series of actual images, and assume that the image with the largest figure of merit is the "best" image. In this regard reference is made, for example, to F.C.A. Groen, I.T. Young, G. Lighthart, "A comparison of different focus functions for use in autofocus algorithms", Cytometry, Vol. 6, pgs. 81-91 (1985). Typically, the quality and effectiveness of such an algorithm is tested by comparing the image selected using the algorithm to an image selected by a human as the "best" image. In this regard, reference can be made to, for example, any of the following publications (and the above-mentione...

AI Technical Summary

Problems solved by technology

As such, the conventional approach to evaluating image processing algorithms outlined in Figure 1A is inappropriate and error-prone

Images