Use of image analysis for sludge characterisation: studying the relation between floc shape and sludge settleability.

This paper starts by presenting a fully automatic image analysis procedure for characterisation of flocs and filaments in activated sludge images. Thereafter the attention is directed towards the results of four lab-scale experiments, in which image information is related to sludge settleability in terms of sludge volume index. This relation is statistically confirmed by applying a principal component analysis to the data. In addition, the redundancy in the data sets is studied with regard to floc shape descriptors and the monitoring potential of image analysis is demonstrated by means of a multiple linear regression exercise.

Activated sludge image analysis system: monitoring settleability and effluent clarity.

A fully automized image analysis procedure is presented for fast and reliable characterization of the activated sludge composition. In previous research, relations between filament abundance and floc shape on the one hand, and the Sludge Volume Index on the other hand were sought. This work aims at demonstrating the use of the image analysis information as an indication for the amount of suspended solids in the effluent.

Evaluation of different shape parameters to distinguish between flocs and filaments in activated sludge images.

The ratio of flocs to filaments in activated sludge waste water treatment plants is of extreme importance for the overall performance of the plant. In order to control this ratio the individual concentrations of flocs and filaments need to be measurable. However, no sensors which can measure these concentrations are currently available. It is proposed that by means of image analysis techniques the ratio of flocs to filaments can be determined. Combination of this ratio with the total biomass concentration results in the individual floc and filament concentration. This contribution focuses on the last step of the image analysis procedure, i.e., the classification of objects as either floc or filament. Five different shape parameters, i.e., aspect ratio, roundness, form factor, fractal dimension and reduced radius of gyration, are evaluated and compared. The results indicate that the form factor is the least suitable and the reduced radius of gyration the most suitable shape parameter to accurately classify flocs and filaments in activated sludge images.

Monitoring activated sludge settling properties using image analysis.

The goal of this study is to develop a monitoring system for activated sludge properties, as this is an essential tool in the battle against filamentous bulking. A fully automatic image analysis procedure for recognising and characterising flocs and filaments in activated sludge images has been optimised and subsequently used to monitor activated sludge properties in a lab-scale installation. The results of two experiments indicate that the image information correlates well with the Sludge Volume Index. It is shown that, at the onset of filamentous bulking, there is an increase in total filament length on the one hand, and a change in floc shape on the other hand.

On the development of a novel image analysis technique to distinguish between flocs and filaments in activated sludge images.

The ratio of flocs to filaments in activated sludge wastewater treatment plants is of extreme importance for the overall performance of the plant. In order to control this ratio the individual concentrations of flocs and filaments need to be measurable. However, no sensors which can measure these concentrations are currently available. In this paper it is outlined how a distinction can be made between flocs and filaments by means of image analysis techniques. Combination of this information with the total biomass concentration results in the individual floc and filament concentrations. The distinction of objects of interest from the background is a crucial step in the image analysis procedure. An automatic thresholding algorithm is proposed which selects two thresholds in images with one fraction darker than the background and the other fraction brighter than the background. Once the objects are separated from the background, they are classified as either flocs or filaments by means of the reduced radius of gyration.

Predicting the onset of filamentous bulking in biological wastewater treatment systems by exploiting image analysis information.

The performance of the activated sludge process is limited by the ability of the sedimentation tank (1) to separate the activated sludge from the treated effluent and (2) to concentrate it. Apart from bad operating strategies or poorly designed clarifiers, settling failures can mainly be attributed to filamentous bulking. Image analysis is a promising technique that can be used for early detection of filamentous bulking. The aim of this paper is therefore twofold. Foremost, correlations are sought between image analysis information (i.e., the total filament length per image, the mean form factor, the mean equivalent floc diameter, the mean floc roundness and the mean floc reduced radius of gyration) and classical measurements (i.e., the Sludge Volume Index (SVI)). Secondly, this information is both explored and exploited in order to identify dynamic ARX and state space-type models. Their performance is compared based on two criteria.

Towards on-line quantification of flocs and filaments by means of image analysis for optimization and control of activated sludge plants.

One of the main reasons for failing of the sedimentation process in activated sludge waste water treatment systems is filamentous bulking. This is a problem of world-wide nature that occurs when the ratio of filamentous to floc forming bacteria is too large. A fully automatic image analysis method for recognizing flocs and filaments in an activated sludge sample is presented. Computer generated images of flocs and filaments, as well as images of activated sludge, are used to assess the capability of simple shape descriptors to distinguish between flocs and filaments. Five different shape descriptors are scrutinized, three of which are derived from size descriptors, namely, the aspect ratio, the roundness, and the form factor. The fractal dimension and the reduced radius of gyration complete the list. The results yield that the form factor is the least suitable and the reduced radius of gyration is the most suitable shape descriptor to accurately identify flocs and filaments in a random activated sludge sample. Based on this, the ratio of flocs to filaments can be estimated automatically. Furthermore, it is proposed that by combining this information with a total biomass concentration measurement, the individual concentrations of flocs and filaments can be determined. Consequently, the method presented may form the basis for a sensor that acts as an early warning system for filamentous bulking.

Image analysis as a monitoring tool for activated sludge properties in lab-scale installations.

An important step in the battle against filamentous bulking is the development of a monitoring system for activated sludge properties. Therefore, a fully automatic image analysis method for recognizing and characterizing flocs and filaments in activated sludge images has been developed. This procedure has been subsequently used to monitor activated sludge properties in a lab-scale installation. The results of a 100-days experiment indicate that the image information correlates well with the evolution of standard settling properties, in this case the Sludge Volume Index. It is shown that, at the onset of severe filamentous bulking, there is an increase in total filament length on the one hand, and a significant change in floc shape on the other hand.