An Improved Detection Method for Hyperspectral Imagery Based on White Gaussian Noise

To solve the low detection efficiency of the present hyperspectral detection method based on adaptive coherence estimator (ACE), an improved detection method based on white Gaussian noise (WGN) is proposed in this paper. Primarily the method uses the spectral angle mapping (SAM) method to adaptively set an optimal signal-to-noise (SNR) parameter based on the hyperspectral image. Then, a corresponding white Gaussian noise is generated according to this SNR parameter and is added to the original image to get a new image data. Finally, based on the new image data, a better target detection result can be obtained by using the ACE detection algorithm. The image data, added to the white Gaussian noise, are more consistent with the theoretical hypotheses of the ACE algorithm. Therefore the detection performance of the algorithm can be efficiently improved. Meanwhile, the adaptivity of setting the optimum SNR parameter in various images can make the method more universal. Experimental results of real world hyperspectral data show that the proposed ACE-WGN method can effectively improve detection performance.