Analysis of Liquid Viscosity by Image Processing Techniques

Objective: In this paper, a technique is proposed for measurement of viscosity using the principle of dispersion of incident light with variation of liquid viscosity. The objective of the proposed technique is to design a non contact technique to analyze the characteristics of dispersed image using image processing for the light incident on liquid whose viscosity is to be measured. Methods: A camera is used to capture the background image of the refracted image. This captured image is processed by techniques like thresholding, filtering, and histogram to arrive at a quantified relation between viscosity and histogram values. To establish the relation an artificial neural network model is designed. Proposed neural network is trained by Levenburg Marquardt Algorithm, once trained it is tested with the real time system conditions. Findings: Once the neural network is designed, tests are conducted. Several samples of liquid are used with varying viscosity. From the obtained results it is clear that the proposed technique is able to measure viscosity accurately with a ischolar_main mean percentage of error 2.01%. Application: From the obtained results it is clear that the proposed viscosity measurement technique can be used for measurement of liquid viscosity, even in dynamic flow conditions.