Modeling, simulation, and analysis of optical remote sensing systems

John Paul Kerekes, Purdue University

Abstract

Remote Sensing of the Earth's resources has evolved from a scientific experiment to a commonly used technological tool. The scientific applications and engineering aspects of remote sensing systems have been extensively studied. However, most of these studies have been aimed at understanding individual components of the remote sensing process, with relatively few studying their interrelationships. A motivation for studying these interrelations has arisen with the advent of user configurable sensors. It will be increasingly necessary for users to understand the tradeoffs of system parameters. In this thesis, two approaches to investigating remote sensing systems are developed. In one approach, detailed models of the scene, the sensor, and the processing aspects of the system are implemented in a discrete simulation. This approach is useful in creating simulated images with desired characteristics for use in sensor or processing algorithm development. An alternative method based on a analytical model of the system is also developed. In this model the informational classes are parameterized by their spectral mean and covariance. These class statistics are modified by models for the system components and an estimate made of the classification accuracy among the informational classes. Application of these models is made to the study of the High Resolution Imaging Spectrometer (HIRIS). The interrelations among system parameters are investigated with several interesting results. Reduced classification accuracy in hazy atmospheres is seen to be due not only to sensor noise, but also to the increased path radiance scattered from the surface. In clear atmospheres, increasing the view angle is seen to result in an increase in classification accuracy due to reduced scene variation as the ground size of image pixels is increased. However, in hazy atmospheres the reduced transmittance and increased path radiance counter this effect and increasing view angle results in decreased accuracy. The relationship between the Signal-to-Noise Ratio (SNR) and classification accuracy is seen to depend in a complex manner on spatial parameters and feature selection. Higher SNR values are seen to not always result in higher accuracies, and even in cases of low SNR feature sets chosen appropriately can lead to high accuracies.

Degree

Ph.D.

Advisors

Landgrebe, Purdue University.

Subject Area

Electrical engineering|Remote sensing

COinS