Method for determining region boundaries for transient data comparison using Feature Selective Validation (FSV)

Data to be compared, such as transients, which is naturally segmented presents a particular set of problems for data comparison and validation in general. Two of the most challenging problems are (a) how to segment and weight the data so that the aspects that dominate the visual assessment also dominate the computed results; (b) how to ensure that the original data is not ‘engineered’ to give a biased result by lengthening or shortening the ‘tails’. Previous studies have looked at performing this segmentation. However, only an average location of the boundaries was determined but there was a clear range of opinions being expressed by the expert users. This paper sets out an algorithmic approach to determining those pre- and post- transient regions, which depends only on the data and requires no visual intervention: thus providing an objective and repeatable method for determining those boundaries. The transition boundaries identified visually have some variation in their location, this paper presents a method to account for these interpretational differences and applies a Monte Carlo solution to provide a ‘quality distribution’ of the comparisons.