The passive remote chemical plume quantification problem may be approached from multiple aspects, corresponding to a variety of physical effects that may be exploited. Accordingly, a diversity of statistical quantification algorithms has been proposed in the literature. The ultimate performance and algorithmic complexity of each is influenced by the assumptions made about the scene, which may include the presence of ancillary measurements or particular background/plume features that may or may not be present. In this work, we evaluate and investigate the advantages and limitations of a number of quantification algorithms that span a variety of such assumptions. With these in-depth insights we gain, a new quantification algorithm is proposed for single gas quantification which is superior to all state-of-the-art algorithms in every almost every aspects including applicability, accuracy, and efficiency. The new method, called selected-band algorithm, achieves its superior performance through an accurate estimation of the unobservable off-plume radiance. The reason why off-plume radiance is recoverable relies on a common observation that most chemical gases only exhibit strong absorptive behavior in certain spectral bands. Those spectral bands where the gas absorption is almost zero or small are ideal to carry out background estimation. In this thesis, the new selected-band algorithm is first derived from its favorable narrow-band sharp-featured gas and then extended to an iterative algorithm that suits all kinds of gases. The performance improvement is verified by simulated data for a variety of experimental settings.
展开▼
