This paper deals with subpixel target detection algorithm NSAR-NPAMF based on spectral smoothing in hyperspectral remote sensing images.Firstly, several typical algorithms of full-pixel target detection, subpixel target detection and NSAR-NPAMF are introduced in this paper.Secondly, based on analysis of limitations of typical target detection algorithms in actual appliactions and spectral disadvantages of hyperspectral data which effect the detection results, three algorithms of NSAR-NPAMF based on spectral smoothing are implemented here.Finally, ROC curve evaluation, target-background separation figure, background feature evaluation and histogram evaluation are introduced and the spectral-smoothing- based NSAR-NPAMF algorithms are quantificationally compared with the original NSAR-NPAMF algorithm by these performance evaluation methods. |