Artificial vision through the correlation coefficient for retinoscopy examinations

Abstract

An automated system was designed and implemented to perform retinoscopy examinations using artificial vision with the correlation coefficient for the processing of retinal images. The technique of observation and experimentation was used to identify the necessary conditions involved in performing the retinoscopy examination. A mechanical, electronic, and artificial vision system design was established to select and install the necessary components for automation. For the operation of the automated system, an algorithm for processing and comparison of images programmed in the LabVIEW software was developed; through a connection to the serial port, data is sent from the computer to the Arduino Uno development board, which controls actuators that are servomotors that direct the position of the lenses and effect the movement of the light beam of the retinoscope. For the automatic change of lenses, a suitable mechanical system was implemented to obtain a fixed retina image; a human-machine interface (HMI) was created, allowing the operator to view the examination process in real-time. With the implementation of the system, the average time of the image comparison was determined, with the result that eyes without ametropia were compared in 162,28 ms, myopia in 166,02 ms, with farsightedness in 170,7 ms and with astigmatism in 174,75 ms, reducing the time taken to perform the retinoscopy examination of 250 to 150 seconds equivalent to 40%. It was concluded that the time of comparison of the images increased by 2.44% while the sea was the refractive damage.