Automatic Defect Recognition using Full Matrix Capture data

This paper describes the development of an automatic defect recognition system applicable to FMC imaged data.  Computer vision principles were used on FMC reconstructed images for feature extraction, and combined with a multilayer perceptron artificial neural network for classification.  A wide variety of single v-weld training samples was used to train the artificial network, which was then tested to determine accuracy.  Automatic defect classification of real single v-weld inspections reported a high level of success.  By training an artificial neural network with information relating to defect orientation, size and location extracted automatically using computer vision techniques is was shown at with little or no user intervention automated defect classification was possible.  The ability to automatically determine defect characteristics offers significant advantages for NDT.