An alternative cost-effective image processing based sensor for continuous turbidity monitoring

Turbidity is the degree to which the optical clarity of water is reduced by impurities in the water. High turbidity values in rivers and lakes promote the growth of pathogen, reduce dissolved oxygen levels and reduce light penetration. The conventional ways of on-site turbidity measurements involve the use of optical sensors similar to those used in commercial turbidimeters. However, these instruments require frequent maintenance due to biological fouling on the sensors. Thus, image processing was proposed as an alternative technique for continuous turbidity measurement to reduce frequency of maintenance. The camera was kept out of water to avoid biofouling while other parts of the system submerged in water can be coated with anti-fouling surface. The setup developed consisting of a webcam, a light source, a microprocessor and a motor used to control the depth of a reference object. The image processing algorithm quantifies the relationship between the number of circles detected on the reference object and the depth of the reference object. By relating the quantified data to turbidity, the setup was able to detect turbidity levels from 20 NTU to 380 NTU with measurement error of 15.7 percent. The repeatability and sensitivity of the turbidity measurement was found to be satisfactory.