Feature analysis of activated sludge based on microscopic images

Activated sludge is a dense liquid containing large quantities of microorganisms, which are capable of neutralizing most organic pollutants. It is also one of the most important techniques for treating municipal and industrial wastewater. Microscopic analysis of activated sludge can serve as a valuable source of information about its condition. In this paper, we describe image processing techniques used to perform detection of flocks and filamentous bacteria colonies on gray scale microscopic images of activated sludge. Two approaches are demonstrated and compared: separate detections with the use variance and Laplacian of Gaussian operators and joint detection based on a fractal dimension operator. It is demonstrated that the fractal dimension is particularly useful for performing texture-based image segmentation.