Date of Award

3-2023

Document Type

Thesis

Degree Name

Master of Science

Department

Department of Electrical and Computer Engineering

First Advisor

Richard K. Martin, PhD

Abstract

An end-to-end LADAR system is modeled at the waveform level to perform material classification at a per-pixel basis. A K-Nearest Neighbors machine learning algorithm is chosen to make predictions using polarimetric material characteristics as features. A variable receiver design is modeled to allow for the use of multiple configurations of Polarization State Analyzers. This research investigates the inclusion of multiple wavelengths in the transmitted laser pulse to improve classification accuracy. Additionally, the effects of lowering the receiver’s detector bandwidth are investigated. Through the classification process, transmitting a multispectral laser pulse is shown to improve classification and may improve future LADAR performance.

AFIT Designator

AFIT-ENG-MS-23-M-043

Comments

A 12-month embargo was observed.

Approved for public release: 88ABW-2023-0404

Download

Share

COinS