Advanced Anomalous Pixel Correction Algorithms for Hyperspectral Thermal Infrared Data: The TASI-600 Case Study

Abstract

Anomalous pixel responses often seriously affect remote sensing applications, especially in the thermal spectral range. In this paper, a new method to identify and correct anomalous pixel responses is presented. The method was specifically developed to handle with hyperspectral data and is based on the statistical analysis of a gray scale RX detector (RXD) image applied on the focal plane space rather than on the image space. An iterative thresholding method to correct anomalous pixels in automatic modality was tuned. Moreover, a band depth-based method to properly restore the lost information was applied. The band depth method serves to prevent the creation of new artifacts during the anomalous pixel correction that could affect applications such as anomaly or change detection and classification for thermal infrared (TIR) hyperspectral imagery. In this paper, we take into consideration hyperspectral TASI-600 data acquired during recent airborne campaigns in Europe. Evidences of the benefits on remote sensing applications such as classification and change detection algorithms in urban areas are shown.