Metrics
58 File views/ downloads
66 Record Views
Thesis
Traversable Region Identification from Post-Disaster Aerial Footage Using Graph Based Image Segmentation
Master of Science (MS), University of West Florida
Spring 2022
Abstract
This research effort proposes a novel method for identifying traversable regions (extracting roads) from aerial images taken after a disaster using graph based image segmentation. The dataset that is used consists of images taken by a UAV at the University of West Florida after hurricane Sally. New ground truth masks were created for these images which divide the image pixels into three categories: road, non-road, and uncertain. A specific preprocessing step is implemented which first uses Catmull-Rom cubic interpolation to resize the image, the Gaussian filter used in Efficient Graph Based Image Segmentation is replaced with a fast median filter, and the color space is converted from RGB to HSV. The Efficient Graph Based Image Segmentation is modified by: changing the Moore pixel neighborhood to the Von Neumann pixel neighborhood, introducing a new adaptive isoperimetric quotient threshold function, changing the distance function used to create the graph edges, changing the sorting algorithm so that the algorithm can run faster, and finally adding a simple function to automatically compute the k (scale) parameter. A new post processing heuristic is proposed for road extraction. The first phase of this heuristic is called median color quantization, which allows for segments to be represented by the median color of the pixels making up the segment. A road segment similarity metric is introduced which is then used to find the segment with the highest likelihood of being road. This segment is called the nucleation site, from which all other road segments are found by performing a recursive hue and saturation comparison of all neighboring segments with the nucleation site. This post processing heuristic helps address the problem of having to use the minimum, rather than some quantile such as the median, edge weight to find evidence of a boundary between segments in Efficient Graph Based image segmentation. The Intersection over Union evaluation metric is used to quantify the road extraction performance. The proposed method maintains high performance on all of the images in the dataset and achieves an Intersection over Union (IoU) score of over double a similar road extraction technique using K-means clustering.
Files and links (1)
CC BY V4.0, Open Access
Details
- Title
- Traversable Region Identification from Post-Disaster Aerial Footage Using Graph Based Image Segmentation
- Resource Type
- Thesis
- Contributors
- Hakki E Sevil (Advisor)Raid Amin (Committee Member)Achraf Cohen (Committee Member)
- Number of pages
- 102
- Identifiers
- 99380090296506600
- Academic Unit
- Hal Marcus College of Science and Engineering
- Language
- English
- Awarding Institution
- University of West Florida; Master of Science (MS)
- Theses and Dissertations
- Master of Science (MS), University of West Florida