Studying the Performance of Object Recognition with Fusion of Visible Light and Infrared Images with Neural Networks

<p>Neural networks have been used for object detection and recognition in both color and intensity camera images. As the use of infrared cameras, colloquially termed thermal cameras, has increased and costs have decreased, object detection and recognition in infrared camera images have been increasingly studied. An infrared image is treated as an intensity image, just like a grayscale camera image, except the intensity corresponds to infrared radiation instead of visible light. The information provided by these two types of images are different, especially in different lighting and environmental situations, and some types of objects are more easily recognized in visible light images, and other types are more easily recognized in infrared images. The fusion of information from both types of images can lead to better object recognition outcomes. Here, several fusion methods of infrared images and visible light images are examined, including a Hue Saturation Value-based fusion method that has not been evaluated in effectiveness in object recognition. This study evaluates the feasibility of recognizing objects with these fused visible light-infrared images across autonomous systems, including commercial, military and law enforcement applications.</p>

Name thesis:degree_name

Master of Science in Mechanical Engineering

Level thesis:degree_level

Thesis - Open Access

Discipline thesis:degree_discipline

Mechanical Engineering

Year

2024

Author dc:creator

• Petkov, Plamen

• Infrared
• Camera
• Machine Learning
• Object Recognition
• Electrical and Computer Engineering
• Mechanical Engineering
• Robotics

Repository record dc:identifier

https://commons.erau.edu/edt/837

OAI identifier oai:identifier

oai:commons.erau.edu:edt-1868