Video-Based Microsleep Detection Using Eyelid Position Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current microsleep detection systems rely on indirect measures of driver drowsiness and often fail to provide timely warnings, as they can be ignored by drivers and do not accurately assess instantaneous eye state.
Innovation Solution
A method utilizing video-based driver observation and lid opening detection to read eye closure information, classify eyelid positions, and calculate a sleep recognition value to identify microsleep before it occurs, incorporating eye movement and gaze direction analysis to output warnings.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If video-based driver observation and lid opening detection are used, then measurement precision of eye state is improved, but device complexity increases
Solution Approach 1:
The patent replaces complex mechanical or sensor-based eye state detection systems with a video-based optical detection system. The method uses standard video cameras to capture eye region images and processes these images through image analysis algorithms to determine eyelid position and eye state, thereby achieving high measurement precision while avoiding the complexity of specialized mechanical sensors or multiple complex detection devices.
2Reliability
If instantaneous eye state classification is implemented, then reliability of microsleep detection is improved, but loss of time for data processing increases
Solution Approach 1:
The system performs preliminary classification of eye states by continuously analyzing video frames to determine eyelid position and eye opening degree before microsleep occurs. By establishing baseline eye state patterns and detecting deviations in real-time, the system prepares detection data in advance, enabling rapid and reliable microsleep identification without requiring extensive post-processing time.
Solution Approach 2:
The patent implements efficient image processing that skips unnecessary computational steps by focusing analysis only on the eye region of video frames. The system directly extracts eyelid position information from video images through targeted image processing algorithms, bypassing full-frame analysis and reducing processing time while maintaining detection reliability.
3Productivity
If eye closure information reading and classification are performed continuously, then productivity of detection system is improved, but use of energy increases
Solution Approach 1:
The system performs eye state classification at periodic intervals rather than continuously processing every video frame. By analyzing eye region images at optimized time intervals and using motion detection to trigger analysis only when eye state changes are detected, the system maintains high detection productivity while significantly reducing energy consumption compared to continuous full-frame processing.
Data Source
AI summary
A method for recognizing microsleep on the part of a driver of a vehicle. The method includes at least a step of reading in an eye closure information item regarding an eye parameter of the driver, the eye closure information item representing a first eyelid position for a maximum eye opening level, and/or a second eyelid position for a minimum eye opening level, for the driver; a step of classifying a current eyelid position of an eyelid of the driver using the eye closure information item, in order to obtain an eye opening information item that represents an open state of the eyes or a closed state of the eyes; and a step of ascertaining a sleep recognition value that represents an indication of an occurrence of microsleep.