Purpose
High-speed imaging is required to capture the rapid eyelid motion during blinking. However, the minimum frame rate necessary for accurate measurement remains unclear. This study investigated how video frame rate affects the accuracy of blink metrics.
Methods
Spontaneous blinks were recorded in 20 participants and reflex blinks were recorded in an additional 10 participants using a 500 frames per second (fps) high-speed infrared camera. Videos were downsampled to simulate acquisition at 250, 100, 50 and 25 fps. Blink metrics were extracted at each frame rate: average and maximum closing/opening velocities, closed-phase duration, total duration, and amplitude. Percentage differences from the 500-fps reference were calculated and agreement was assessed using Bland-Altman analysis with limits of agreement (LoA).
Results
Agreement with the 500-fps reference varied across metrics. Blink amplitude and duration showed minimal bias and narrow LoA across all frame rates. In contrast, velocity metrics were underestimated at low frame rates, particularly for maximum velocities and reflex blinks. For spontaneous blinks, average closing velocity showed good agreement at 100–250 fps, whereas for reflex blinks, only 250 fps achieved comparable measurements. Average opening velocity was less sensitive to frame rate, with ≥50 fps showing minimal bias. Maximum velocities were markedly underestimated below 250 fps for both blink types.
Conclusion
Video frame rate significantly impacts the accuracy of certain blink metrics. Low frame rates may be sufficient for estimating blink amplitude and duration, while velocity metrics require higher sampling. A minimum of 250 fps is recommended for a comprehensive assessment of blink dynamics.
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号