SURFACE ILLUMINATION LEVEL ANALYSIS BASED ON COMPUTER INFORMATION PROCESSING METHODS USING MACHINE LEARNING

Buryi Anton Sergeyevich

The article is devoted to the analysis of the surface illumination level based on computer methods of information processing and machine learning. A convolutional neural network model has been developed that is capable of determining illumination from images taken at fixed shooting parameters. It has been found that the use of logarithmic transformation and the error backpropagation algorithm reduces the impact of outliers and increases the accuracy of the model, while data optimization and reduction of the illumination range (no more than 600 lux, no more than 400 lux and no more than 300 lux) ensure the stability of the results. The scientific novelty lies in the adaptation of machine learning methods for analyzing illumination at the microlevel, which expands the possibilities of automated illumination control.

Keywords:illumination, convolutional neural network, machine learning, computer analysis, automation