Options for Solid Point Target Detection in Hyperspectral Data

Adaptive target detection algorithms need to estimate parameters before the actual calculation of the metric for target detection. For solid subpixel targets in hyperspectral data, we consider two algorithms for effective detection. The first algorithm uses explicitly only background estimation; the second estimates both the background and the target fraction in the pixel. We compare the performance of each algorithm; we also consider the need to alter the estimated covariance matrix based on the estimated target fraction.