Far Infrared Detection Algorithms for Vulnerable Road Users Protection

Far Infrared (FIR) technology was originated a few decades ago in the military domain for its ability to detect any dissipation of heat, as this reveals not only the presence of moving threats or vehicles, but also of living creatures, including human beings, in a passive way, by daytime or night time, even in adverse conditions. More recently, FIR was found a valuable sensor to detect Vulnerable Road Users (VRU) in the automotive safety domain. After a first adaptation to VRU detection of typical FIR detection techniques used in the military domain, where most applications tend to rather concentrate on small objects at a great distance, we have also investigated more recent theoretical schemes, applying them to Far Infrared images. Candidate schemes include various learning methods, such as neural networks, support vector machine, boosting.