Using Image Quality Assessment (IQA) Databases to Provide an Appraisal of the Ability of the Feature Selective Validation Method (FSV) to Compare Two-Dimensional Datasets

This paper investigates the strengths and drawbacks of the recently developed feature selective validation (FSV)-2D method. Considering that a subjective benchmark for the validation of two-dimensional computational electromagnetics data is not available, five datasets with subjective scores, commonly used in image quality assessment, are used. It is found that the FSV-2D prediction is influenced by image type and distortion type. Encouraged by the assessment results, eight parameters of the FSV-2D method are optimized by use of genetic algorithms. It is shown that the optimized FSV-2D method provides better correlation with subjective scores. Good agreement with theoretical analysis for computational electromagnetic data further validates the proposed approach.