Publication | Closed Access
DEVELOPMENT OF A MANUAL CRACK QUANTIFICATION AND AUTOMATED CRACK MEASUREMENT SYSTEM
12
Citations
0
References
2005
Year
EngineeringMechanical EngineeringImage DatabaseDamage MechanismImage AnalysisData SciencePattern RecognitionVerify OutputsEdge DetectionMachine VisionStructural Health MonitoringCrack Type IndexComputer ScienceMedical Image ComputingUnified Crack IndexOptical Image RecognitionAutomated InspectionComputer VisionCivil EngineeringRemote SensingTexture AnalysisCrack FormationDynamic Crack PropagationMechanics Of MaterialsFracture Mechanics
The objective of this research was to (1) develop a manual image analysis system (MIAS), (2) verify outputs provided by a consultant, and (3) develop an automated image analysis system (AIAS). The MIAS was designed in consultation with Iowa Department of Transportation personnel. It allows them to process digital images interactively through a computer screen in the most efficient manner. For each image, the manually processed data are stored automatically in the database to verify the outputs provided by the consultant. A total of 1,585 images from the consultant were processed using the MIAS. A robust tile-based automated crack imaging software package was developed, which can compute unified crack index and crack type index. First, the image is overlaid with a tile of predetermined dimensions, such as 6 in. x 6 in. of a pavement surface. To improve the accuracy of the automated crack imaging procedure, a variable optimum threshold level was then applied for classifying a tile as cracked or not. Using 10 sample images, the AIAS result was compared against the one produced by MIAS resulting in 3.4% precision and -3.4% bias.