Protótipo de turbidímetro de baixo custo com processamento de imagem

Abstract

This work presents the development of a low-cost optical turbidimeter designed to estimate water turbidity through digital image processing techniques. The prototype consists of an ESP32-S3 microcontroller, an OV2640 camera without an infrared filter, and an infrared laser beam of 808 nm, aligned in a compact, sealed structure made of stainless steel tubing and 3D-printed parts. The method is based on capturing two images per sample — one with the laser turned off and another with the laser turned on — and applying digital subtraction between them to isolate the light beam and minimize external interferences. From the resulting image, a column-wise average intensity vector was generated, and its area under the curve was used as an explanatory variable for linear regression modeling. The system was calibrated with a reference multiparameter probe (Hanna HI9829), achieving a coefficient of determination 𝑅2 = 0.9987, MAE of 2.30FNU, and RMSE of 3.48FNU, demonstrating a high correlation between predicted and measured values. The results indicate that the proposed approach is a technically feasible, robust, and economically accessible alternative for applications in environmental monitoring, industrial processes, and educational activities, standing out for using image sensors instead of conventional photodetectors.