Ph.D. Defense: Runchen Zhao
Ph.D. Thesis Defense
LWIR Spectral Variability Integration and Improvements to an Earth Observing Statistical Performance Model
Runchen Zhao
Imaging Science PhD Candidate
Chester F. Carlson Center for Imaging Science, RIT
Abstract: Remote sensing systems can be used to identify objects without physical contact. In hyperspectral remote sensing systems, pixels from a 3D “hypercube” can be described as a one-dimensional spectrum. This one-dimensional spectrum contains information about surface reflectance, emissivity, and variation, as a function of wavelength, and is often thought of as the materials spectral signature or fingerprint. This spectral signature can be used in applications such as spectral target detection. To study the spectral signatures variation and its impact on target detection, a parameter trade-off study is often needed. In this research, a full-spectrum spectral imaging system analytical model, called the Forecasting and Analysis of Spectroradiometric System Performance (FASSP), has been used and improved upon. FASSP uses first- and second-order statistics of surface spectral reflectances or emissivities and temperature variations to describe a user-defined scenario. The statistics are propagated through an imaging system-like model with the addition of atmosphere effects and sensor noise, for example. The system can then use this information in a target detection application where results are illustrated in the form of a Receiver Operating Characteristic (ROC) curve. The current FASSP model performs target detection in both the VNIR and LWIR. More specifically, in the LWIR the model relied on at-sensor radiance statistics. However, in most realistic cases, target detection in the LWIR is applied to retrieved emissivities where surface temperature is also a key obtained parameter. This retrieved domain was absent in the FASSP model. Therefore, to obtain retrieved surface temperatures and simulate more realistic LWIR scenarios, a statistical approach to temperature/emissivity separation (TES) called the statistical iterative spectrally smooth temperature-emissivity separation (S-ISSTES) algorithm has been derived and integrated into the FASSP model. The new S-ISSTES module can retrieve first- and second-order statistics of surface emissivities and ground temperatures. This work was derived and validated in a study that used HyTES LWIR data. Additionally, an equivalent adaptive cosine estimator (ACE) detector called the cotangent detector (COT) was constructed for integration into the FASSP model, which was also validated. Lastly, we constructed a study using the Hapke mixing model, based on user defined particle-related parameters coupled with Monte-Carlo simulations, to generate needed covariance matrices for mixed particle material detection studies using FASSP.
Intended Audience: Undergraduates, graduates, and experts. Those with interest in the topic.
To request an interpreter, please visit https://myaccess.rit.edu
Event Snapshot
When and Where
Who
Open to the Public
Interpreter Requested?
No