Numerical Tests-Based Assessment of the Procedures of Blurry and Noisy Image Enhancement

Abstract

An extension of the test-based method of multi-aspect numerical assessment of the quality of image enhancement procedures on blurry and noisy images is presented. The principles of construction of the tests affected by various types and intensity of blurring and noising are described. Linear models of blurring have been defined. Formal definitions of measurable parameters characterizing the quality of image enhancement procedures are proposed. Comments on the errors and applicability limitations of the proposed testing method are also given. In a numerical experiment, the utility of the method by comparative assessment of 18 various image enhancement procedures is proven. Remarks concerning the properties of several widely-known image enhancement procedures are formulated. Concluding remarks about the utility of the presented method of image enhancement procedures assessment are given.