Background characterization for subpixel target detection Conference Paper uri icon

abstract

  • When performing point target detection in hyperspectral imagery, one often uses the spectral inverse covariance matrix to whiten the natural noise of the image. Since the cube is not necessarily stationary, we wish to understand when segmentation is worthwhile to provide different covariance matrices for different areas of the cube. Using simulations and several new analytical tools, we propose general guidelines for when segmentation is useful.

publication date

  • January 1, 2017