Dynamic Threshold Eye Closure Detection for Driver Monitoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing closed eye determination systems fail to accurately differentiate between eye opening and closure due to fluctuations in eye opening degrees, leading to incorrect detection of eye closure.
Innovation Solution
A closed eye determination device that dynamically sets a threshold value based on local minimum and maximum eye opening values over a certain period, using a processor to detect eye opening changes and adjust the threshold accordingly, with consideration for sleepiness levels and individual variations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a fixed threshold value is used for closed eye determination, then the determination process is simple, but detection accuracy deteriorates due to fluctuations in eye opening degree
Solution Approach 1:
The patent applies dynamics by transitioning from a fixed threshold to a dynamically adjusted threshold that adapts to individual eye characteristics. The system extracts local minimum values from time-series eye opening data and uses these to set personalized reference lower limit values, allowing the threshold to evolve with the driver's actual eye behavior patterns.
Solution Approach 2:
The patent implements parameter changes by modifying the threshold value based on extracted eye opening characteristics. Specifically, the closed eye determination threshold is set as a reference lower limit value plus a predetermined offset, where the reference lower limit itself is derived from statistical analysis of local minimum values in the eye opening time series, thereby adapting the parameter to individual variations.
2Reliability
If a fixed threshold value is used for closed eye determination, then the system operation is simple, but reliability deteriorates due to false detection
Solution Approach 1:
The patent changes the determination parameter from a universal fixed threshold to an adaptive threshold based on individual eye characteristics. By extracting local minimum values from each driver's eye opening time series and calculating reference lower limit values, the system creates personalized thresholds that reliably distinguish between normal eye fluctuations and actual closed-eye states.
Solution Approach 2:
The system incorporates feedback by continuously monitoring eye opening degree and using this information to establish accurate reference values. The time-series analysis of eye opening data provides feedback on individual blinking patterns, which is then used to adjust the closed eye determination threshold, creating a closed-loop system that improves reliability through adaptive learning.
3Measurement precision
If individual variations in eye opening degree are not considered, then the determination process is simple, but detection precision deteriorates
Solution Approach 1:
The patent applies local quality by creating personalized determination parameters for each driver based on their individual eye characteristics. Instead of applying a uniform threshold, the system extracts local minimum values specific to each driver's blinking pattern and establishes reference lower limit values tailored to their unique eye opening degree variations, thereby achieving precise detection adapted to local (individual) characteristics.
Solution Approach 2:
The system performs preliminary action by pre-establishing reference lower limit values through analysis of eye opening time series data before actual closed-eye determination begins. This preliminary characterization of individual eye behavior patterns enables more accurate subsequent detection, as the personalized thresholds are prepared in advance based on observed blinking characteristics.
Data Source
AI summary
A closed eye determination device includes an eye opening degree detection unit that detects a degree of eye opening of a driver from each of a series of face images of a driver in time series in a latest certain period of time being acquired by a driver monitor camera, a local minimum value extraction unit that extracts a local minimum value of the degree of eye opening from a change in time series of the degree of eye opening in the certain period of time, a reference lower limit value-setting unit that sets, as a reference lower limit value of the degree of eye opening, a local minimum value in order corresponding to a blinking number of the driver in the certain period of time in order from the smallest local minimum value among local minimum values of the degree of eye opening in the certain period of time, a threshold value setting unit that sets a closed eye determination threshold value, based on the reference lower limit value, and a closed eye determination unit that determines that an eye of the driver is closed when the degree of eye opening detected from an image acquired by the driver monitor camera after the certain period of time is less than or equal to the closed eye determination threshold value.