DESIGN & IMPLEMENTATION OF COLOR DETECTION SYSTEM IN SELF-STOPPING TITRATOR FOR COD TITRATION USING RASPBERRY PI

• Main Author: Fauzi, Sulhan
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/43569
• Institution: Institut Teknologi Bandung

Water quality testing has several parameters that has to be measured, one of which is the COD (Chemical Oxygen Demand) parameter. Titration is one but many methods and also the most common method to measure COD levels in water. Commonly, the process of titration carried out by visual method (seeing with naked eye) and sometimes gives less acurate results due to wrong detection colors that emerge when end-point is achieved. To deal with this problem, computer vision is use to detect color of liquid at the end-point and measure it’s color representation in HSV values. System color detection works by taking liquid samples pictures at certain framerates and only process some part of the image, image that inside ROI (Region of Interest). Each color represented in frame inside ROI then clustered into some class using K-means and the most dominant color class is use to approach real liquid color at time the frames taken. System is placed under a box called reaction chamber to eliminate some light noises from environment and includes its own lighting to provide best color detection to camera. All system is programmed using python, OpenCV library and runs on Raspberry Pi 3B+. Data from experiment shows that system can stops COD titration at time as it should be and can reached up to 8,06% error and only 0,06% at detecting color, compared to corelDraw color inspection. Nevertheless, system needs more testing with more precise equipment in order to determine system color detection quality in such a responsibly trusted and quantitative result.