Development of comprehensive accuracy assessment indexes for building footprint extraction

Presents a suite of indexes for comprehensively evaluating the results of automated building extraction. The indexes described include detection rate, correctness, matched overlay, area omission error, area commission error, root mean square error, corner difference, area difference, perimeter difference, and shape similarity. These proposed unbiased quality measures should enable the accuracy assessment of the building extraction process to address extraction issues such as completeness, geometric accuracy, and building shape similarity.