Research on Vision Measurement Method Based on Extreme Learning Machine and Distance Detection

An objective of this study was to propose a dynamic adaptive eyesight test method based on extreme learning machine and mobile phone distance detection technology. One of the main obstacles in vision measurement is accuracy and convenience which are antithetical. In this paper an eyesight test APP was developed focused on daily use based on the characteristics of the family environment, which improved the convenience and accuracy of eyesight test on phone. In order to measure the distance from eyes to the phone we adopted the Monocular Vision Technology with the facility of the front camera of the mobile phone. Detecting the brightness of the mobile phone screen and ambient light, considering the congestion and appearance of the visual target, with distance measured before, the APP may calculate user’s eyesight based on the relationship between distance, brightness, congestion of the visual target and the visual acuity value specified by the Near Visual Acuity Chart. Through Extreme Learning Machine, the APP revised the measurement deviation making the measurement result close to the medical optometry result. Based on the method proposed in this paper, a vision detection system was developed and tested. The results show that the average error of the system is 2.80%, and the maximum error is 6.82%.