Progress on the self-service kiosk for testing the UV protection on sunglasses: polynomial and neural network approximation for calculating light transmittance

A method using different light sources and sensors have already been used to approximate weighting functions to calculate light transmittance in sunglasses. Although it made possible a low cost equipment that inform the user about its sunglasses, each transmittance test is still dependent of its components. We tested two methods, using polynomial approximation and artificial neural network, that would open the possibility for the use of a fixed light source and sensor for all light transmittance tests from the standard. Spectrophotometry, visible transmittance and traffic light transmittance was calculated in 45 lenses of sunglasses, used as samples for testing the methodologies. The tests included a white LED, a RGB sensor, and electronic for control and signal acquisition. Bland - Altman analysis tool was used to calculate the agreement between the method and the transmittances calculated in the spectrophotometer. Both methods, had an approximation within the deviation limit required by NBR15111. The system with the polynomial regression showed lower deviations than artificial neural networks. A larger number of samples can improve the methods in order to obtain an optimal calibration that includes all sunglasses. No meter in the market can calculate accurately all light transmittances measurements required for the sunglasses. The methodology was applied only for the visible light, while UV and infrared spectrum remains to be tested. The methodology tested presented a way for simple low-cost equipment for all light transmittance tests in sunglasses