Advanced Image Processing for the Study of Photoactivated Sludge in Wastewater Treatment

INTRODUCTION

Photoactivated sludge is a consortium of microorganisms such as bacteria, microalgae, and protozoa and represents an alternative to conventional wastewater treatment systems, such as activated sludge. Photoactivated sludge significantly reduces the energy consumption of wastewater treatment processes because the oxygen required for biological decontamination processes is provided by the microalgae via photosynthesis and not by a mechanical aeration system. Furthermore, the algal biomass produced, rich in nutrients, organic matter, nitrogen, and phosphorus, can be used as biofuel or fertilizer. However, despite satisfactory treatment performance, certain biological and physicochemical mechanisms that can lead to a loss of effluent quality are still poorly understood.

This internship will be developed as part of the BAP-TI-STE project, an French acronym for Bacteria-Algae-Protozoa, Image Processing for the Wastewater Treatment Plant. The objective of this project is to gain a better understanding of the sedimentation and nitrogen and phosphorus assimilation processes in photoactivated sludge through qualitative, and quantitative visual characterization of the microorganisms present in the sludge. This characterization is a research topic that we will address using optical imaging (macro and microscopic) coupled with image processing, which will utilize machine learning concepts.

The first phase of the BAP-TI-STE project has already been completed. During this phase, a protocol for imaging of photoactivated sludge was developed, and a large number of microscopic (Figure 1A and Figure 2) and macroscopic (Figure 1B) images were produced. These images show more-or-less compact "spots" ranging in size from 0.05 to 2 mm, combining organic matter and bacteria, surrounded by filamentous algae (Figure 1A right) and protozoa (Figure 2). These complex structures are called "flocs." Bacteria can also be observed in these images, dispersed in the media surrounding various flocs.

The images are of sufficient quality to be subject to automatic processing. Images such as those in Figure 1B have already been analyzed using conventional image processing methods: binarization, morphological analysis, size, shape, and counting. However, these methods can have certain limitations, particularly when identifying microorganism species and characterizing complex morphologies.

The second phase of the BAPTISTE project, addressed by this internship offer, aims to improve the processing algorithms to extract as much quantitative information as possible from the images produced in the first phase of the project.

INTERNSHIP TASKS

The intern will be specifically required to:

• Provide an explanatory summary of advanced image processing methods, particularly those based on machine learning.
• Propose an image processing algorithm for the morphological characterization of structures present in the macroscopic and microscopic images obtained during the first phase of the BAPTISTE project.
• Propose an algorithm for identifying and counting the largest microorganisms in flocs, such as microalgae and protozoa.
• Draw comparisons between the results obtained using conventional processing methods and those using machine learning.

• Highlight possible limitations and avenues for improvement of the image acquisition method.