To support emergency first responders after a natural disaster or hazard event, actionable information on the condition of critical infrastructure needs to be generated quickly. The objective of this study is to determine the most effective approach for the detection and delineation of fine-scale crack damage to bridge and road surfaces using rapid, semi-automated image analysis methods. The effectiveness of these image analysis methods are tested using both single-date imagery and multiple-date image pairs in an attempt to validate the benefits associated with the collection of a baseline imagery catalog needed to employ multiple-date, change detection approaches. Tests of various pixel-based, object-based and spatial contextual methods are performed to determine the ideal combination of operations, parameters and thresholds resulting in the highest detection rate that simultaneously minimizes false detections. Methods employing a multiple-date, change detection approach are shown to outperform those techniques relying on single-date imagery only. With a producer’s accuracy of 77.9% and user’s accuracy of 69.1%, a pixel-based image difference model using kernel filters to perform spatial filtering operations is the most successful
