DVS Eye Blink Detection for Low-Light Cabin Fatigue Monitoring
Solution Overview
Problem
Conventional cameras struggle to accurately detect eye blink patterns due to the small size of the eye and fast blink duration, especially in low-light conditions, and high-speed cameras increase cost and data processing demands.
Innovation Solution
An eye blink detection method using a DVS camera that integrates asynchronous DVS pixels to form frames, identifying patterns where a first color region and a second color region are distributed one above the other in the eye region to determine eye blink actions, calculating average height differences to confirm these patterns.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional cameras are used for eye blink detection, then the device cost and data processing load are reduced, but the measurement precision of eye blink patterns deteriorates due to the small eye area and fast blink duration
Solution Approach 1:
The patent changes the fundamental parameter of camera operation from frame-based capture to event-based capture. DVS cameras operate by detecting changes in light intensity at each pixel independently and asynchronously, rather than capturing complete frames at fixed intervals. This parameter change enables high temporal resolution for detecting fast eye blink events without requiring high-speed conventional cameras, thus improving measurement precision while avoiding the associated complexity and cost.
2Measurement precision
If high-speed cameras with frame rates exceeding 100 frames per second are used, then the measurement precision of eye blink detection is improved, but the device cost and data processing amount significantly increase
Solution Approach 1:
The patent extracts only the relevant information (light intensity changes at specific pixel locations) from the continuous visual scene, rather than capturing and processing complete high-resolution frames at high speeds. The DVS camera outputs only pixel events where light intensity changes occur, filtering out redundant information and reducing data processing requirements while maintaining detection precision.
3Measurement precision
If conventional cameras are used in low-light conditions, then the device cost is reduced, but the measurement precision of eye blink detection deteriorates due to difficulty in capturing clear images
Solution Approach 1:
The patent changes the detection parameter from capturing absolute light intensity levels to detecting relative light intensity changes. Since eye blinks cause localized changes in reflected light from the eye region, the DVS camera can detect these changes even in low-light conditions by focusing on the delta (change) rather than the absolute intensity, thereby improving measurement precision independent of ambient illumination levels.
Data Source
AI summary
An eye blink detection method and system are disclosed. The eye blink detection method comprises: photographing a face using a DVS camera to obtain a stream of DVS pixels; integrating DVS pixels of the stream of DVS pixels to form a plurality of DVS frames, wherein each of the plurality of DVS frames comprises a plurality of first and second color pixels, each being associated with one or more DVS pixels indicating a brightening event and each of the second color pixel being associated with one or more DVS pixels indicating a darkening event; and determining whether there exists an eye blink action in a DVS frame of the plurality of DVS frames, wherein the determining comprises: determining whether there exists a pattern in which a first and second color region are distributed one above the other in an eye region of the at least one DVS frame.



