Water Utilization Monitoring System by using Image Processing

Abstract

Water scarcity is a big problem in dry season around the world. Many countries are trying to manage and minimize the use of water. Water management is essential to help reduce this problem. For example,   to regularly check the integrity of the plumbing system to prevent damage. Therefore, the authors have an idea for applying image processing technology to check the water meter page. Meter images were prepared and eliminated noises by a Gaussian filter. Then, the filtered images were binarized from color image by Otsu’s algorithm. Next, the morphological was applied to extract the region on interest (ROI). Finally, the Python-tesseract which is an optical character recognition (OCR) tool was implemented to convert string to text(numbers). After that the results will be sent via WIFI to be stored and displayed in graphs on the web page. If there is any abnormality in water consumption data from the meter reading, the system will send notification emails to the relevant department.  The experimental results shown that the average percentage of recognition accuracy for each image types are 48%, 75.1%, and 91.64% respectively.