LITIV

 

LITIV (Laboratoire d'Interprétation et de Traitement d'Images et Vidéo) is a computer vision laboratory studying image and video processing. More specifically, LITIV focuses in the processing of visible and infrared videos to detect and analyze objects of interest under two complementary modalities. The laboratory is also interested in video analysis in infrared and visible in medical applications. This laboratory is under the responsibility of the professor Guillaume-Alexandre Bilodeau.

 

LITIV PROFILE

The general objective of LITIV is to design methods and algorithms to interpret the action occurring in a scene. The informations of interest are the interactions between humans and the interactions between humans and the objects in their environment. The use of visible cameras allows identifying some interactions in some circumstances, but in the case of similar colors between the scene and human in it, the analysis becomes quite difficult. The LITIV proposes to use infrared cameras to introduce thermal information in the analysis of the scene.

Our research projects thus aim to integrate information from visible and infrared cameras to improve the detection of humans and objects, along with their interactions.

A second topic of interest of LITIV is tracking with (pan-tilt-zoom) PTZ cameras. The interest in using such a camera is a larger coverage of a scene with a single camera. Controlling the camera to perform accurate tracking is the challenge that we are undertaking.

Finally, LITIV is also interested in the contribution of infrared cameras to medical experiments. Infrared, more specifically thermography, allows obtaining temperature measurements in real-time of the body of a subject. Our project aims to improve the measurements.

RESEARCH INTERESTS

• Video surveillance
• Visible/infrared registration
• Object tracking
• Visible/infrared stereoscopy
• Thermographic image analysis
• Medical applications of computer vision
• Content-based image retrieval
• Object recognition
• Figure/ground segmentation
• Object modeling

LITIV Wordle

 

Laboratory member of:

FUNDING: