Localized processing for hyperspectral image analysis

Qian Du

doi:10.1117/12.570784

Translator Disclaimer

14 December 2004 Localized processing for hyperspectral image analysis

Qian Du

Author Affiliations +

Proceedings Volume 5584, Chemical and Biological Standoff Detection II; (2004) https://doi.org/10.1117/12.570784
Event: Optics East, 2004, Philadelphia, Pennsylvania, United States

Abstract

Target detection is one of the major tasks in hyperspectral image analysis. Constrained Energy Minimization (CEM) is a popular technique for target detection. It designs a finite impulse response filter in such a manner that the filter output energy is minimized subject to a constraint imposed by the desired target of interest. It is particular useful when only the desired target signature is available. When those undesired signatures to be eliminated are also known, Target Constrained Interference Minimization Filter (TCIMF) can be used to minimize the output of undesired signatures to further improve the performance. It has been demonstrated that TCIMF can better differentiate targets with similar spectral signatures. Both CEM and TCIMF involve the calculation of the data sample correlation matrix R and its inverse matrix R^-1. The function of R^-1 is background suppression. When the target to be detected is very small and embedded at the sub-pixel level, it is difficult to detect it. But if the data sample correlation matrix R can well present the statistics of the background surrounding the pixel containing the object such that R^-1 can well suppress the background, the target may still have a chance to be detected. So in this paper we propose a localized processing technique. Instead of using all the pixels in an image scene to calculate the R, only several lines of pixels near the pixel to be processed are used for the R computation. The preliminary result using an HYDICE image scene demonstrates the effectiveness of such a localized processing technique in the detection of targets at sub-pixel level. Interestingly, in some cases it can also improve the performance of CEM in target discrimination.

Citation Download Citation

Qian Du "Localized processing for hyperspectral image analysis", Proc. SPIE 5584, Chemical and Biological Standoff Detection II, (14 December 2004); https://doi.org/10.1117/12.570784

ACCESS THE FULL ARTICLE