Study on rapid colorimetric detection of pH and free chlorine by multi-view feature extraction and CNN fusion

To address the issues of significant subjective errors and high susceptibility to interference from environmental factors and application scenarios in traditional colorimetry, and digital colorimetry for detecting pH and free chlorine. The present study innovatively proposes a detection method. This integrating digital image colorimetry with convolutional neural network (CNN). A portable, accurate, and efficient multi-angle rapid detection platform was developed, and the visualization of the detection process and results was realized via an APP with a built-in CNN model. Compared with the traditional fixed-angle detection method, multi-angle rotation detection effectively reduces the maximum detection error. For water sample detection within pH range of 6–9 and a free chlorine range of 0–5 ppm, the maximum detection error was 0.06 for pH and 0.09 ppm for free chlorine, with a response time of 1.2 s for both. The regression coefficients reached 0.9989 and 0.9987, respectively. With low detection errors, this method enables the rapid and accurate detection of water pH value and free chlorine content. This study provides new insights and a theoretical basis for the design and development of water quality detection equipment, and offers guiding significance for promoting technological upgrading and equipment innovation in the field of water quality detection.