The Use of Neural Network Coupled with Image Processing for Water Quality Assessment (Location: Hot Spring Mae-Khachan, Thailand)

Abstract

Parameters such as pH, alkalinity, total hardness, and nitrate are important to determine the quality of agricultural purpose. Water quality testing strips are popular for testing the water. In addition to the low price, this tool is very easy to use. However, bias due to colour reading from different users exists. In this study, the evaluation of water quality with image processing coupled with the artificial neural network (ANN) model is proposed. Water quality is assessed using testing strips. Images of the strip are taken by mobile-phones and geo-tag provided by GPS. Then, colour-space transformation, image enhancement, and regions of interest (ROI) are carried out as part of the image processing stage. Finally, the ANN-multi layer perceptron (MLP) is used to obtain water quality predictions based on the results of image analysis. Based on the results of the analysis, the R2 values > 0.80 for the estimating water parameters. This research shows that the image processing coupled with ANN has the potential to get a more precise estimation value for water quality assessment. For future work, the geo-tagging application meets with the data centre system in this study offers periodic monitoring of water quality in a large area.