United States Department of Transportation
i
Image Processing and Machine Learning Techniques for Automated Detection of Planes at Utah Airports
-
2021-06-01
[PDF-4.18 MB]
Details:
-
Creators:
-
Corporate Creators:
-
Corporate Contributors:
-
Subject/TRT Terms:
-
Publication/ Report Number:
-
Resource Type:
-
Geographical Coverage:
-
Edition:Final Report August 2020 to November 2021
-
Corporate Publisher:
-
Abstract:Most of the airports in the United States are non-towered airports. Utah is not an exception to this rule. As a result, these airports fall behind in terms of aircraft operation count and identification. On the other hand, image detection and recognition have long been assisting different industrial areas in shifting towards automation in performing numerous tasks. That said, this work attempts to utilize various image processing and machine learning techniques to automatically detect and count the airplanes in operation at the Utah airports. The first major task to accomplish such an automatic system is data collection. Data collection will be divided into office and in-field data collections. The office data collection aims to build a bridge from identified aircraft at the airports to FAA-registered aircraft data. Also, a great amount of in-field data collection is needed first to find a solution for an optimal camera deployment at airports and second to create a data repository of different aircraft transported through the airports. This data includes aircraft trajectory while in operation on or near the airport runway environment as well as the image data captured from the operating aircraft. The former helps us detect the strategic airfield placement for camera deployments for accurate aircraft operation detection. The latter is required for feeding the machine learning computer packages as training data. Data processing is the second major task and structures the software development of the project. This task comprises several required subtasks before finalizing an aircraft operation count and identification. These subtasks include but are not limited to establishing vision-based algorithms for motion detection, aircraft detection, operation tracking, aircraft trajectory quantification, tail number region detection, tail number character recognition, and aircraft operation identification. This report also comprehensively reviews the computer vision methods tested to accomplish the above-mentioned subtasks.
-
Format:
-
Funding:
-
Collection(s):
-
Main Document Checksum:
-
Download URL:
-
File Type:
