We consider a multi-channel remote field eddy current sensor apparatus, which is installed on a mobile robot deployed in pipelines with the mission of detecting defects. Features in raw sensory data that are associated with defects could be masked by noise and therefore difficult to identify in some instances. In order to enhance these features that potentially identify defects, we propose an entropy filter that maps raw sensory data points into a local entropy measure. In the entropy space, data are then classified by means of a thresholding procedure based on the Neyman–Pearson criterion. The effectiveness of the algorithm is demonstrated by applying it to different data sets obtained from field trials.
展开▼
