Semester of Graduation
Master of Science in Electrical Engineering (MSEE)
Division of Electrical and Computer Engineering
In this thesis work, we study the visible-light and near-infrared images for practical applications. Different colormaps, contours, or pixel-based indicators can be extracted from these optical images. Among them, normalized difference vegetation index (NDVI) has been widely utilized for applications in vegetation-health monitoring, geological survey, and climate forecast, etc. recently. We extend the use of NDVI for the navigation over the uncharted geographical areas. Henceforth, we propose a new algorithm which is called multi-resolution Dijkstra algorithm to find the approximate shortest paths within the large planar graphs representing the real terrains.
The outline of this thesis is briefed as follows. The relevant literature review is given in Chapter 1. The basics of NDVI are introduced in Chapter 2. In Chapter 3, for a given NDVI colormap, we design a new algorithm to convert it to a black- and-white colormap, which can specify the passable and inhibited region(s). Such a black-and-white colormap can be easily translated to a large planar graph if every pixel is considered a vertex or node. To navigate through a large planar graph, the conventional Dijkstra algorithm may not converge or would take forever to converge. Therefore, we introduce our proposed novel multi-resolution Dijkstra algorithm in Chapter 4, which can deal with the navigation in a much more efficient manner. The theoretical analysis, namely worst-scenario analysis, is established in Chapter 5 for our proposed algorithm. The concluding remarks will be drawn in Chapter 6.
Tian, Rui, "New Multi-Resolution Shortest-Path Navigation Techniques Using the NDVI Imagery" (2017). LSU Master's Theses. 4360.
Available for download on Tuesday, November 12, 2024