Vehicle bottom anomaly detection algorithm based on SIFT

To achieve anomaly detection of vehicle bottom images, this paper studied several extraction algorithms of local invariant image feature signal, the SIFT (Scale Invariant Feature Transform) algorithm was improved and the feasibility was verified by the vehicle bottom image anomaly detection with the improved algorithm. This algorithm automatically created and extracted features, set the threshold then searched for matching feature points, removed duplicate matching feature points, obtained rotation angle and scaling, revised the image to locate anomalies by cross-correlation, then marked and displayed abnormal sites. This algorithm established and extracted feature point fast, matched feature points accurately, abnormal positions marked clearly. (C) 2015 Elsevier GmbH. All rights reserved.