Research And Application Of Blink Detection Method Based On Pixel Fluctuation Ratio Of Eye Image

Nowadays,with the rapid development and application of computer science and artificial intelligence,people’s facial information has attracted more and more attention from scholars in the field of artificial intelligence.Among them,eyes are the most important part of facial information,and the detection and feature study of eyes is an important research direction in the field of human-computer interaction technology and artificial intelligence.The combination of human eye detection technology and human-computer interaction technology has played an important role in many fields,such as fatigue driving,eye tracking and so on.At present,some of the existing blink detection technologies need expensive high-precision cameras and infrared cameras to achieve this requirement.Although high-precision equipment can meet the detection needs,the cost is too high,and some of the equipment may be potentially dangerous to the body.So developing a low-cost,secure and easy to implement system is where we need to go.To solve these problems,a new blink recognition algorithm is proposed in this paper.The main research results of this paper are as follows:1.We combine the latest facial recognition algorithm technology with image processing technology to propose a blink detection algorithm as FRP algorithm that can be completed with ordinary camera.This algorithm is based on the pixel fluctuation ratio(FRP)of the eye region image for blink detection.According to face sign algorithm to locate the eye area,then converts the image to a grayscale image,according to the gray image in the eye area to detect the point,the eye area from eye to eye end is divided into three sections,calculating the value of pixels in each paragraph,and calculating the part of the image pixels is worth fluctuations than(FRP),using support vector machine(SVM)for the selected training sample training.We tested it on two data sets,and the blink detection algorithm achieved good results on both data sets.2.The pixel fluctuation ratio of the gray image of the eyes was used to identify whether the subjects had single eyelids or double eyelids as as SDFRP algorithm.This algorithm locates the eye region according to the face recognition technology,converts the image of this region into grayscale image,frames the eyelid region according to the detected registration points,selects two lines,calculates the pixel fluctuation ratio of each line on these two lines,and then takes the average.The calculated fluctuation ratio was imported into the SVM trainer for training.If the comparison value was within a certain range,it was judged as a single eyelid;if the comparison value was beyond this range,it was judged as a double eyelid.3.Based on the blink recognition algorithm proposed in this paper,a word input system based on low cost blink recognition is designed.The blink recognition algorithm is combined with word input to complete a simple human-computer interaction system.Good results have been obtained after testing.