Occupant posture analysis with stereo and thermal infrared video: algorithms and experimental evaluation

Dynamic analysis of vehicle occupant posture is a key requirement in designing "smart airbag" systems. Vision-based technology could enable the use of precise information about the occupant's size, posture, and, in particular, position in making airbag-deployment decisions. Novel sensory systems and algorithms need to be developed for capture, analysis, and classification of dynamic video-based information for a new generation of safe airbags. This paper presents a systematic investigation in which stereo and thermal long-wavelength infrared video-based real-time vision systems are developed and systematically evaluated. It also includes the design of several test beds, including instrumented vehicles for systematic experimental studies for the evaluation of independent and comparative evaluation in automobiles. Results of extensive experimental trials suggest basic feasibility of stereo and thermal long-wavelength infrared video-based occupant position and posture-analysis system.