Conflation System MBP

MBPConflate is proof-of-concept software developed to aid in conflation research. The program implements a powerful algorithm for automatic multi-source image conflation that utilizes new Algebraic Systems-based similarity metrics during the feature comparison process. One benefit of integrating Algebraic Systems into the feature comparison process is that more global shape properties can be discovered, preserved, and used for matching. The algorithm presented in this paper clusters the individual attributes from two data sets based on connectivity and neighboring features. The clustered sets are then compared using a heirarchial rule-based technique that includes measures of structural similarity. User selected criteria define a match between features and the user assists feature matching only when ambiguity cannot be resolved automatically. Experiments have shown that the algorithm is computationally efficient and robust.