A novel endmember extraction and discrimination algorithm for target detection in hyperspectral imagery

作者：He Yuanlei^*; Liu Daizhi; Yi Shihua

来源：Journal of Optics, 2011, 13(8): 085402.

DOI：10.1088/2040-8978/13/8/085402

• hyperspectral imagery
• target detection
• endmember extraction

Target detection is one of the most important applications of hyperspectral imagery. In this paper, an endmember extraction and discrimination algorithm (EEDA) was presented for hyperspectral target detection. Unlike most of the existing endmember extraction techniques, the EEDA takes advantage of the FastICA-generated independent components (ICs) that separate all potential endmember pixels in individual components and uses the maximum spectral screening (MSS) to select the best representative background endmembers and thus does not require prior knowledge of the number of endmembers to be extracted. Besides, it has the ability to discriminate the endmember of targets of interest from the interfering background endmembers by identifying the IC that contains the most target information. In order to demonstrate the utility of the EEDA, the fully constrained least-squares (FCLS) algorithm is implemented to estimate the target abundance fractions of the image pixels. Experimental results on the airborne visible/infrared imaging spectrometer (AVIRIS) dataset show that the proposed EEDA in conjunction with the FCLS yields better detection performance compared with two well-known target detectors, an adaptive cosine estimator (ACE) and an adaptive matched filter (AMF).

• 出版日期2011-8