Toronto Metropolitan University
Browse
- No file added yet -

Performance Analysis of Multispectral LiDAR in Land Cover Classification

Download (2.88 MB)
thesis
posted on 2023-04-05, 19:10 authored by Khakan Zulfiquar
The Optech Titan is the world’s first multispectral airborne Light Detection and Ranging (LIDAR) sensor, a revolutionary sensor that includes three active imaging channels of different wavelengths for day or night mapping of complex environments. Multispectral imagery and monochromatic LIDAR have long existed as independent technologies and both systems have developed workflows to perform land cover classification. This project was undertaken to analyze the performance of Optech Titan’s three active imaging channels and LIDAR attributes in land cover classification. By processing selective parameters through the multispectral image land cover classification process, we can determine the accuracy performance of individual channels and attributes in land cover classification. The outcome of this process will measure the effectiveness of combining LIDAR attributes with multispectral imagery for land cover classification. The test site was a 600m x 600m residential neighbourhood in Oshawa, Ontario captured at point-spacing of 0.5 meter. Multispectral imagery had an overall accuracy result of 77%. The most accurate land cover classification result from our testing was 77.5%. This was produced as a special index scenario by using the three intensities along with the nDSM. It is apparent from the results that the intensity-attribute provides the most useful information in land cover classification. The highest monochromatic LIDAR accuracy result of 70% came from Channel 2 (NIR - 1024 mm). Channel 2’s accuracy is only 7% lower than multispectral imagery result. Channel 1 and 3 had less-favorable results at 59.5% and 58% respectively. Individual land cover classification tests on Z-attribute and N-attribute produced unfavorable results of 37% and 47.5% respectively.

History

Language

English

Degree

  • Master of Engineering

Program

  • Civil Engineering

Granting Institution

Ryerson University

LAC Thesis Type

  • Thesis

Year

2017

Usage metrics

    Civil Engineering (Theses)

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC