Compulsory internship

Background:
Mixing and homogeneity in large vessels, such as in biogas digesters and wastewater oxidation basins, determines stability, efficiency and productivity of the processes. However, achieving efficient mixing of the huge reactor volumes is a big challenge, since there is a lag of monitoring methods to characterize the flow patterns inside the closed vessels.

Objective:

The project focusses on the development of image processing algorithms for surface flow tracking. Sample images of typical flow behavior are captured by a conventional camera system continuously. These data sets shall be used to develop the tracking algorithms for surface flow, floating layer and homogeneity preferably with available software toolboxes, e.g. imagej, U-net, octave or others.

Tasks:

• Survey on flow imaging toolboxes
• Definition of target parameters for data extraction
• Screening of available sample data sets
• Implementation of imaging processing and data extraction routines
• Documentation and presentation of results

Studies in the area of computer science, informatics, electrical, mechanical engineering or similar

• Comprehensive knowledge of digital image analyses (optical flow, pattern recognition)
• Basics of machine learning based image processing (e.g. U-net)
• Independent and structured way of working