Workgroup Efficient Embedded Systems Hochschule Augsburg
University of Applied Sciences
HS Augsburg / Dept. of Computer Science / Workgroup EES german
Home
Team
Job vacancies
Open
degree theses
Lab
Research
ChASA
ASTERICS
ParaNut
Triokulus
e4rat
HiCoVec
Teaching
Hardware Systems
Development of
Digital Systems 2
Development of
Integrated Systems
Project Works
Publications
Archive
Seminar WS15/16
Seminar WS14/15
Seminar WS13/14
Imprint

The ASTERICS Framework

An Open Toolbox for Sophisticated FPGA-based Image Processing

Image processing on embedded platforms is a challenging task, especially when implementing extensive computer vision applications. Field-programmable gate arrays (FPGAs) offer a suitable technology to accelerate image processing by customized hardware.
The ASTERICS ("Augsburg Sophisticated Toolbox for Embedded Real-time Image Crunching Systems") framework is designed as a modular building set to perform various real-time image processing tasks. It offers modules and interfaces to perform various image processing operations. These range from (a) pixel-based (thresholding, contrast adjustment, ...) and (b) window-oriented filter operations (noise filters, edge enhancement, ...) to more complex, higher-level operations, which are likely to be defined in software. This way, the framework also covers the integration of complex, higher-level (c) semi-global and (d) global algorithms (e.g. needed for object recognition applications). Several high-level algorithms have been adopted and integrated into the frameworks image processing flow, such as undistortion and rectification, natural feature description and edge detection and Hough transform.

Due to the ASTERICS frameworks open structure in terms of flexible data types and the extensibility of the module library, it is an ideal platform to build systems for sophisticated image processing tasks.

Exemplary structure of an ASTERICS-based image processing system:

ASTERICS-based image processing system

Further Information Article on the project (embedded world conference 2015)
Wiki-Pages with Demo Code
22.2.2018 - Michael Schäferling