Research And Implementation Of Embedded Eye-Computer Interactive System

As an important research direction in the field of human-computer interaction,eye-computer interaction technology has been widely used in the fields of medical care,assisted driving,virtual reality and military.The traditional desktop-based eye-computer interaction system needs to rely on too many hardware devices,is not easy to wear,and cannot be used as an auxiliary system for field combat equipment.Therefore,in order to free the warrior’s hands to concentrate on the operation of weapons and combat equipment,it is very significant to study and design a small,lightweight,easily integrated and wearable eye-computer interaction system to improve the intelligence of combat equipment.This paper combines embedded technology and image processing technology to study the embedded eye-computer interaction system.,The main work is as follows:(1)In response to the problems of large size,low portability and limited application scenarios of traditional eye-computer interaction systems,the overall design of the embedded eye-computer interaction system is proposed.RK3399 chip used as the main processor,a USB eye behavior monitoring camera supporting UVC protocol is selected,and the circuit design of each functional module is completed,Linux operating system used as the system software platform in this system.(2)According to the migration process of embedded Linux operating system,The embedded development environment is built,u-boot and Linux kernel are transplanted,root file system is made,camera driver is configured and Open CV library for image processing is transplanted.(3)A software program for eye image acquisition based on V4L2 architecture is designed,and an algorithm for eye state and behavior detection using image processing technology is proposed.The algorithm first preprocesses the acquired eye images,then detects and identifies the eye states using the eye feature extraction method based on binarized images and the individual differentiation threshold selection method,and finally achieves the detection and identification of blinks and characteristic eye behaviors by the blink detection algorithm with continuous image frame processing.(4)The test platform of the eye-computer interaction system is built,and the network connectivity between the PC,virtual machine and development board was realized first;then experiments were conducted for the proposed eye state detection and recognition algorithm,and a threshold selection method for separating the three eye states was obtained,yielding an average recognition accuracy of 97.43% and an average processing speed of 76.9 FPS;then experiments were conducted for the proposed blink detection algorithm Experiments were then conducted for the proposed blink detection algorithm and several characteristic eye behaviors were classified,and the results indicated that the algorithm was able to recognize characteristic eye behaviors such as long blink,double blink,triple blink,and squint;finally,the eye-control telescope was used as the background to analyze the user’s eye-machine interaction law,and the feasibility of the eye-control telescope interaction behavior design was verified through experiments.