Comparison of Three-Dimensional Datasets by Using the Generalized n-Dimensional ( n-D) Feature Selective Validation (FSV) Technique

Abstract

Automatic methods to evaluate the validity of computational electromagnetics computer modeling and simulations have widespread applications. The feature selective validation (FSV) method is a heuristic technique which has been shown to give a broad agreement with a visual assessment for one-dimensional data. As a heuristic technique, extending the dimensionality is an important target for the improvement and development of FSV. One of the major challenges in the development of n -dimensional ( n -D) FSV is the difficulty of obtaining visual assessment results, since, the visual comparison of three- and higher dimensional data is difficult or even impossible. This paper formulates the comparison of 3-D data based on an established generalized n -D-FSV approach. The performance of the approach is investigated by means of the Laboratory for Image and Video Engineering Video Quality Database which provides subjective scores of 150 distorted videos. A statistical evaluation of the relative performance of FSV and other publicly available full-reference video quality assessment algorithms is presented. Further, parameter tuning is performed to improve the agreement of 3-D FSV results and subjective scores. The proposed approach is finally applied to the self-referenced validation of an electromagnetic simulation model to identify and locate the continuous variation of electric field within a region of space.