Back to search

Rochester Institute of Technology

Hyperspectral Imaging System Model Implementation and Analysis

Abstract

dc:description.abstract

In support of hyperspectral imaging system design and parameter trade-off research, an analytical end-to-end model to simulate the remote sensing system pipeline and to forecast remote sensing system performance has been implemented. It is also being made available to the remote sensing community through a website. Users are able to forecast hyperspectral imaging system performance by defining an observational scenario along with imaging system parameters. For system modeling, the implemented analytical model includes scene, sensor and target characteristics as well as atmospheric features, background spectral reflectance statistics, sensor specifications and target class reflectance statistics. The sensor model has been extended to include the airborne ProspecTIR instrument. To validate the analytical model, experiments were designed and conducted. The predictive system model has been verified by comparing the forecast results to ones obtained using real world data collected during the RIT SHARE 2012 collection. Results include the use of large calibration panels to show the predicted radiance consistent with the collected data. Grass radiance predicted from ground truth reflectance data also compare well with the real world collected data, and an eigenvector analysis also supports the validity of the predictions. Two examples of subpixel target detection scenario are presented. One is to detect subpixel wood yellow painted planks in an asphalt playground, and the other is to detect subpixel green painted wood planks in grass. To validate our system performance, the detection performance are analyzed using receiver operating characteristic (ROC) curves in a comprehensive scenario setting. The predicted ROC result of the yellow planks matches well the ROC derived from collected data. However, the predicted ROC curve of green planks differs from collected data ROC curve. Additional experiments were conducted and analyzed to discuss the possible reasons of the mismatch including scene characterization inaccuracy. Several subpixel target detection parameter trade-off analyses are given, including relative calibration error vs SNR, the relationship among probability of detection, meteorological range, pixel fill factor, relative calibration error and false alarm rate. These trade-off analyses explain the utility of this model for hyperspectral imaging system design and research.

Degree

thesis:*
Name thesis:degree_name
Imaging Science (MS)
Level thesis:degree_level
Thesis
Discipline thesis:degree_discipline
Chester F. Carlson Center for Imaging Science (COS)

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Ding, Bo
Contributors dc:contributor
  • John Kerekes

Subjects

dc:subject × 6

Identifiers

dc:identifier.*
Repository record dc:identifier
https://repository.rit.edu/theses/7672

Chain of custody

source
Harvested from
Rochester Institute of Technology
Base URL
repository.rit.edu/do/oai/
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
citation

Ding, Bo. Hyperspectral Imaging System Model Implementation and Analysis. Thesis thesis, https://repository.rit.edu/theses/7672