Target-Constrained Interference-Minimized Filter for subpixel target detection in hyperspectral imagery

Due to significantly improved high spatial and spectral resolution hyperspectral sensors can now uncover many material substances which cannot be resolved by multispectral sensors. However, this also comes at a price that many unknown and unidentified signal sources, referred to as interferers may, also be extracted unexpectedly. Such interferers generally introduce additional noise effects on target detection and its factor must be taken into account. The problem associated with this interference is challenging because interference is generally unknown in nature and cannot be identified from an image scene. This paper presents a Target-Constrained Interference-Minimized Filter (TCIMF) which does not require to identify interferers, but can minimize the effects caused by interference. In addition, the TCIMF also separates undesired targets from the desired targets so that the TCIMF can eliminate undesired targets while detecting the desired targets and minimizing interfering effects. Most attractively, these three operations can be carried out by a single process in real time implementation.