Abstract: In the anomaly detection of hyperspectral images (HSIs), aiming at the difficulty of distinguishing the abnormal target from the background and the low accuracy of background prediction, a new HSI anomaly detection algorithm based on background sparse reconstruction is proposed. An online dictionary learning method is used to estimate the background spectral dictionary. The estimated background image is sparse reconstructed by the learning dictionary. The estimated background image is subtracted from the origin image to get the residual image. The anomaly detection is achieved by using the local RX detector to traverse the residual image. The effectiveness of the proposed HSI anomaly detection algorithm based on the background sparse reconstruction is illustrated in a series of real-world data experiments.

Key words: anomaly detection, hyperspectral image (HSI), dictionary learning, sparse representation, online learning