Real-time Driving Status Monitoring Algorithm Research And System Design

As for avoiding traffic accidents caused by abnormal driving status, it is significant to develop a reliable real-time driving status monitoring system.This theis is devided into two main parts, algorithm research and system design of real-time driving status monitoring.Algorithm research:The algorithms based on computer vision are concerned, including the interior and exterior visual way. And multi-information fusion technology is used to determine driving status. Interior visual way is the algorithm based on eye movement characteristic. Infrared face images are acquired and then eye movement parameters such as PERCLOS, eye closure duration, gaze direction are extracted to discrimate the driving status. Exterior visual way is the algorithm based on vehicle lane departure level. Front-view and side-view lane images are both acquired and vehicle lane departure will be oberserved in a continous time to detect "S" lane, which can be adopted to determine driving status.System design:According to system requirements analysis, hardware based on TMS320DM6446 ARM+DSP dual-cores processor is designed as well as the software, including software frame design and application software design. ARM is responsible for video capuring, DSP algorithms calling and alarm/video signals outputting, while DSP is responsible for driving status monitoring algorithms implematation. Simulations of embedded real-time driving status monitoring based on eye movement characteristic demonstrated the feasibility of this system.