Image Recognition and Tracking for Use in Augmented Reality Applications
Session Number
F817
Advisor(s)
Benjamin Carls, Fermi National Accelerator Laboratory
Location
B-116
Start Date
28-4-2016 11:05 AM
End Date
28-4-2016 11:30 AM
Abstract
Image recognition software attempts to identify similar or identical images by searching for and matching visual features of two sampled images. However, one issue when it comes to image processing is accounting for inconsistencies in viewing conditions. This investigation explored varying factors that could impact the effectiveness of the image recognition and tracking software, Vuforia, in the game engine Unity. The performance and reliability of detection and tracking of Different image targets for Vuforia were tested. Images that were rich in detail with complex features provided better tracking and detection than simpler images. Complex images can include images of the Chicago skyline and simple images could range from a bare, empty wall to a bookshelf with bland or similarly colored books. Both insufficient and excessive lighting caused image detection to fail to some extent. Images with repeating parts or patterns would fail to be tracked effectively, as the software would have trouble determining the position of the image. Today’s image recognition software still has much to expand on regarding accuracy, efficiency, and flexibility, but image recognition is still able to track an image successfully. Even in largely varying environments, images are still able to be recognized by computer software.
Image Recognition and Tracking for Use in Augmented Reality Applications
B-116
Image recognition software attempts to identify similar or identical images by searching for and matching visual features of two sampled images. However, one issue when it comes to image processing is accounting for inconsistencies in viewing conditions. This investigation explored varying factors that could impact the effectiveness of the image recognition and tracking software, Vuforia, in the game engine Unity. The performance and reliability of detection and tracking of Different image targets for Vuforia were tested. Images that were rich in detail with complex features provided better tracking and detection than simpler images. Complex images can include images of the Chicago skyline and simple images could range from a bare, empty wall to a bookshelf with bland or similarly colored books. Both insufficient and excessive lighting caused image detection to fail to some extent. Images with repeating parts or patterns would fail to be tracked effectively, as the software would have trouble determining the position of the image. Today’s image recognition software still has much to expand on regarding accuracy, efficiency, and flexibility, but image recognition is still able to track an image successfully. Even in largely varying environments, images are still able to be recognized by computer software.