Infrared-Visible video registration

It has been shown in the past, that the combination infrared and visible images improves at short and long distance the detection and segmentation of people, which are the main object of interest in visual surveillance. Indeed, people in visible imagery can be hard to detect when the color contrast with background is low and when the illumination of the scene is poor. In contrast, IR imagery performs generally well in these situations. However, IR imagery is not perfect either as a good thermal contrast is required to detect people. As a result, it is beneficial to use both modalities jointly to capitalize on both of their strength. In our case, we use long wavelength IR and visible cameras in a stereo configuration, which can also provide coarse depth information. We consider both planar and non-planar scenes.
Combining visible and IR to improve the detection of people requires solving many problems. It is especially difficult to find features that match in both modalities to perform registration. Even after registration, both aligned silhouettes (IR/Vis) will be imperfectly segmented. Which parts correspond to the human, which parts do not? My research objectives match these research problems towards the solution for combining IR/Vis human silhouettes for improve detection and tracking of people.

Faculty Supervisor:

Guillaume-Alexandre Bilodeau

Student:

Luan Nguyen

Partner:

Discipline:

Engineering - computer / electrical

Sector:

University:

Program:

Globalink