Computer Researchers Work On Auto SafetyComputer Researchers Work On Auto Safety

Researchers at the University of California, San Diego are developing multimodal computer vision techniques that could be used for surveillance and driver safety.

UCSD professor Mohan Trivedi said the technology uses multiple viewpoints and technologies for interpretation of human movements and activities. Trivedi, who teaches electrical and computer engineering at UCSD's Jacobs School of Engineering, said the system reflected as well as emitted energies.

Computer vision and robotics researchers, backed by Volkswagen-Audi and the UC Discovery program, installed four cameras with infrared and color technology in the LISA-Q, an Infiniti Q45 to track a driver. The video-based system monitored the driver's head, arms, torso and legs for thorough, real-time tracking.

Researchers said they proved the system's effectiveness in "fairly" accurately tracking performance in a real-world situation with noise and various levels of lighting. The system could be used for driver safety and assistance features in smart cars.

"The multi-perspective characteristics of the system provide redundant trajectories of the body parts, while the multimodal characteristics of the system provides robustness and reliability of feature detection and tracking," the researchers wrote in a recent report. "The combination of a deterministic activity grammar (called 'operation triplet') and a Hidden Markov model-based classifier provides semantic-level analysis of human activity."

Researchers also developed an algorithm for finding accurate correspondence between objects viewed by a stereo head with one thermal infrared eye and a second eye that detects color. The algorithms can be applied to multiple objects at multiple depths for surveillance.

"This can lead to robust and accurate pedestrian detection, tracking and analysis for active safety systems in a vehicle, and also for operating surveillance systems on a 24/7 basis," student Steve Krotosky said.