Research On Eye Tracking Coding Servo Control Method For Human-computer Interaction

Aiming at the problems of low accuracy,poor reliability and insufficient human eye intention initiative in the field of eye movement servo control in the field of human-computer visual interaction,a servo control method based on human eye movement coding technology is proposed.The intelligent servo control of human eye intention is realized by encoding and processing human eye movement instructions and integrating them with human eye movement intention.In order to construct the spatial model of eye movement visual servo control,based on the three-dimensional visual field of human vision global machine vision local machine vision,after synchronous dynamic matching of human-computer binocular vision,the tracking field of view real-time switching method is used to accurately collect the data of adult eye gaze coordinates,tracking target coordinates and servo control coordinates.In order to improve the accuracy of eye movement servo control,an eye movement command coding processing method is proposed,which uses the trained Ada Boost fusion strong classifier algorithm and the real-time algorithm in the landmark detector detection video sequence to complete the adult eye location and blink determination.A pupil location method based on improved multi corner joint detection is proposed.The pupil location is completed by using the eye image point information,and the digital coding model of human eye command action is constructed,which provides an instruction library for eye motion servo control.In order to improve the reliability and intention initiative in eye movement servo control,a reinforcement learning method combining eye movement coding instruction and intention is proposed.By combining the improved ddpg continuous control strategy reinforcement learning algorithm with eye movement intention instruction,we can perceive the environmental state and learn the optimal strategy of eye movement instruction selection in the process of human-computer interaction.Through experimental comparison,the accuracy of grasping objects by integrating human eye movement intention reasoning method is about 15% higher than that of conventional methods.