Accession Number:

AD1132718

Title:

Surface Defect Detection in Aircraft Skin and Visual Navigation Based on Forced Feature Selection through Segmentation

Descriptive Note:

[Technical Report, Master's Thesis]

Corporate Author:

AIR FORCE INSTITUTE OF TECHNOLOGY WRIGHT-PATTERSON AFB OH

Personal Author(s):

• Hussey, Tyler B.

Report Date:

2021-03-01

Pagination or Media Count:

117

Abstract:

Visual inspection of aircraft skin for surface defects is an area of maintenance that is particularly intensive for time and manpower. One novel way to combat this problem is through the use of computer vision and the advent of Artificial Neural Networks ANN, or more specifically, semantic segmentation via Convolutional Neural Networks CNN. The research in the paper explores the use of semantic segmentation of aerial imagery as a way to force feature selection onto key areas of an image that might be more likely to correspond under seasonal variations. Utilizing feature selection and matching on the masked aerial image and the satellite image produces a set of reliable key points that can be used for camera pose estimation and visual navigation.

Descriptors:

• computer vision
• neural networks
• artificial intelligence software
• computational science
• supervised machine learning
• artificial intelligence
• machine learning
• aircrafts
• automata theory
• information science
• image recognition
• dimensionality reduction
• network science
• air force
• computers
• data science
• image processing
• data mining
• information systems
• convolutional neural networks

Subject Categories:

• Cybernetics

Distribution Statement:

[A, Approved For Public Release]