Blink oscillogram generation method, device and equipment based on deep learning

Abstract

The invention provides a blink oscillogram generation method, device and equipment based on deep learning, and the method comprises the steps: obtaining a video stream for the eye movement of a user,and enabling the video stream to comprise a plurality of eye image frames; inputting each eye image frame in the plurality of eye image frames into a pre-trained segmentation model, and obtaining a target area contained between the upper eyelid and the lower eyelid in each image frame; obtaining the palpebral fissure height corresponding to the target area; and calculating the opening degree of the palpebral fissure height and the height threshold of each eye image frame according to the image frame sequence of the video stream, and generating a blink oscillogram according to the opening degree sequence. According to the application, the accuracy of inter-eyelid area identification and the accuracy of the palpebral fissure height and the blinking oscillogram can be improved, so that the reliability of evaluating ocular surface diseases is improved.

Description

technical field [0001] The present application relates to the technical field of image processing, and in particular to a method, device and equipment for generating eye blink waveforms based on deep learning. Background technique [0002] When the patient's ocular surface is abnormal, there may be varying degrees of burning sensation, irritation or foreign body sensation in the eye, resulting in changes in blink frequency, abnormal blink amplitude, and abnormal complete closure time. Ocular surface disease can be assessed by drawing eye blink waveforms. [0003] At present, when drawing the blink waveform, the edge detection algorithm is usually used to capture the lid margin information from the collected video frames. In this solution, because the edge of the upper eyelid is cut off by useless edges such as eyelashes, it is impossible to accurately locate the edge of the upper eyelid. position, which leads to the inability to accurately calculate the eyelid distance, and...

AI Technical Summary

Problems solved by technology

[0003] At present, when drawing the blink waveform, the edge detection algorithm is usually used to capture the lid margin information from the collected video frames. In this solution, because the edge of the upper eyelid is cut off by useless edges such as eyelashes, it is impossible to accurately locate the edge of the upper eyelid. position, which leads to the inability to accurately calculate the eyelid distance, and the accuracy of the blink waveform needs to be improved

Images