IMPROVED HYPERSPECTRAL ANOMALY TARGET DETECTION METHOD BASED ON MEAN VALUE ADJUSTMENT

With the development of sensors' capability, there has been an increasing interest in target detection using hyperspectral imagery. As the classical algorithms for hyperspectral imagery anomaly target detection, the Reed-Xiaoli detector (RXD) and the low probability target detector (LPTD) algorithms cannot describe the complex background very well. Therefore, the detection results of the RXD and LPTD algorithms maybe have a high false alarm rate under a certain detection rate. In this paper, an improved hyperspectral anomaly target detection method based on mean value adjustment is proposed to reduce the false alarm rate. There are three main steps in our proposed method: 1) traditional anomaly detection; 2) dividing the detected image into the background and the potential area of the anomaly target according to the preliminary detection results; 3) comparing the similarity between each potential area pixel and the mean value of the whole image, and determining the final outcome. Experiments with both synthetic and real hyperspectral data sets indicate that the improved method could reduce the false alarm rate and improve detection performance effectively compared with original algorithms.