Segmentation Approach Towards Phase-Contrast Microscopic Images of Activated Sludge to Monitor the Wastewater...

Area of Science:

• Environmental Science
• Microbiology
• Image Analysis

Background:

• Activated sludge (AS) wastewater treatment relies on microbial flocs and filamentous bacteria.
• Effective monitoring and fault diagnosis are crucial for plant efficiency.
• Phase-contrast microscopic (PCM) images are vital for AS analysis but present artifacts.

Purpose of the Study:

• To propose and assess nine distinct image segmentation strategies for AS samples.
• To evaluate the effectiveness of each method considering PCM artifacts.
• To identify optimal segmentation approaches for AS image analysis.

Main Methods:

• Nine image segmentation algorithms were applied to PCM images of AS.
• Methods included color space analysis, edge detection, clustering, adaptive thresholding, texture analysis, watershed, split-and-merge, Kittler's thresholding, and filtering.
• Segmentation performance was critically analyzed against PCM artifacts.
• Gold standard ground truth images were used for quantitative assessment.

Main Results:

• Edge detection-based segmentation demonstrated superior accuracy.
• Texture-based segmentation yielded the lowest false negative ratio.
• Different segmentation methods showed varying suitability depending on specific analysis needs and artifact presence.

Conclusions:

• Edge detection is highly effective for accurate AS image segmentation.
• Texture-based segmentation is optimal for minimizing missed filamentous bacteria.
• The choice of segmentation method should be tailored to the specific diagnostic goals and image characteristics.