Quantifying Relative Turbidity Levels using Image Processing Techniques

Since the past decade, underwater imaging has become an important area of research as it finds widespread applications in the fields of seafloor exploration, resource mapping etc., with the help of Manned and Unmanned underwater vehicles. But the quality of images captured underwater majorly suffer due to light attenuation and scattering, the latter of which is caused primarily due to turbidity. The main objective of this paper is to quantify the relative turbidity present in the acquired underwater image. In this work a three step image enhancement process, which are threshold, contrast limited adaptive histogram equalization and color balancing are carried out.The third stage output is taken as one of the reference images for estimation of turbidity level of the acquired image. The other extreme is created by extraction of an appropriate region of interest from input image whose pixel intensity is averaged and set as the most turbid image. This model is translated into a working user interface application which is tested using different images acquired in varying light conditions over a wide range of depths up to 3000m to validate the efficacy of the proposed methodology.The proposed technique can be used in underwater control systems for navigation and mapping in the given location, where it is impervious to use an appropriate model that is best suited for the water conditions.