Drowsiness prediction method, device and storage medium

Abstract

The present invention provides a drowsiness prediction method, which includes the following steps: step S100, acquiring an image of the detection target, performing face detection and skin recognition, including: detecting the target facial area and extracting and clustering related skin areas; Part aspect ratio; step S200, the clustering result of each skin area is processed respectively; By comparing the clustering result and the RGB average value corresponding to each frame, extract and calculate the rPPG signal, and finally pass the fast Fourier transform method Estimating the target heart rate; step S300 , selecting the optimal signal detected through different clustering results to estimate the target sleepiness level. The invention can objectively and truly reflect the drowsiness of the driver.

Description

technical field [0001] The invention relates to the technical field of biotechnology, in particular to a drowsiness prediction method and device for drivers. Background technique [0002] The existing method is to judge drowsiness by segmenting the monitoring image and comparing facial features, including the position of key points such as mouth and eyes, and to give early warning. Doppler radar and complex signal processing methods are used to obtain the The tester’s restless emotional activities, blink frequency and duration and other fatigue data are used to judge whether the testee is dozing off or falling asleep; the sleepiness can be judged by changing the positions of key points such as mouth and eyes over time, refer to The elements are relatively single, and cannot objectively and truly reflect the sleepiness of the testee, and are low in technicality, low in accuracy, and lagging. [0003] The term rPPG (remote Photo PlethysmoGraphy) involved in the present invent...

AI Technical Summary

Problems solved by technology

[0002] The existing method is to judge drowsiness by segmenting the monitoring image and comparing facial features, including the position of key points such as mouth and eyes, and to give early warning. Doppler radar and complex signal processing methods are used to obtain the The tester’s restless emotional activities, blink frequency and duration and other fatigue data are used to judge whether the testee is dozing off or falling asleep; the sleepiness can be judged by changing the positions of key points such as mouth and eyes over time, refer to The elements are relatively single, and cannot objectively and truly reflect the sleepiness of the testee, and are low in technicality, low in accuracy, and lagging

Images