The Research Of Non-contact Real-time Eye Recognition System Based On Smart Camera

The state of the human eye and the motion trajectory are called eye movement information which has a close association from psychological characteristics to behavioral characteristics of human beings.Presently,common eye detection equipment are based on PC machine with the testing equipment which to fix the head.This product not only has big limit for its large volume but also cause larger discomfort to experimenter.In recent years,the development of domestic helmet eye recognition systems uses infrared light and sensors.This will lead to the body damage and a further thing that there are speed and accuracy defects.This paper researched and designed the eye movement recognition system and method based on the smart camera,in-depth analyzed the development of eye movement recognition technology and testing principles and characteristics of various eye movement detection systems,proposed a non-contact real-time eye movement recognition system based on a smart camera.Excellent computing ability of embedded systems balances the speed and accuracy to meet the needs of real-time performance.It has small size so that easy to carry and the good security,accuracy and reliability has a good prospect in the future development.First,according to the requirement of the eye movement recognition detection designed the system architecture and selected the suitable camera,image sensors and human-computer interaction equipment.Then,based on the difficulty in the eye recognition,specifically formulated the plan,and determined the software development platform.Second,it researched the image matching algorithm which is used in eye recognition.In order to adapt to natural light during the test and make the eye feature areas and other areas have enough gray scale difference,We should do initial adjustment before inspection according to the different environment.When during the process it should be able to adjust the parameters such as aperture and exposure time,to ensure that there are enough contrast,brightness,and robustness.In the study of real-time eye recognition algorithm,presented repeated-detection algorithm based on structure matching which constructed sequentially lower resolution images and used fast correlation coefficient matching roughly determine the matching region.Then,this area would converted to a layer of the region of interest by coordinate transformation and using the least squares matching in the upper image matching so that the gray scale change,rotation,scaling can be identified.This algorithm can greatly improve the computing speed.Finally,it designed a real-time eye movement feedback program interface on the G language platform and realized the real-time eye movement recognition system based on smart camera and industrial touch screen which display the detected result via Ethernet communications.This paper talked out a series of experiments in variety of common interference and showed the data with analysis.It discussed the accuracy,stability of eye recognition system and the efficiency of the image processing.Experimental results show that the eye recognition system based on smart camera has compact structure,high detection speed and high accuracy results.