Authors:
Dennis Schuldt
1
;
Jörg Thiem
1
and
Semir Mustedanagic
2
Affiliations:
1
University of Applied Sciences and Arts Dortmund, Germany
;
2
Smart Mechatronics GmbH, Germany
Keyword(s):
MATLAB, MATLAB Embedded Coder, ARM Cortex-A8, DSP C674x, VRmagic D3 Platform, Texas Instruments.
Related
Ontology
Subjects/Areas/Topics:
Engineering Applications
;
Image Processing
;
Informatics in Control, Automation and Robotics
;
Intelligent Control Systems and Optimization
;
Robot Design, Development and Control
;
Robotics and Automation
;
Signal Processing, Sensors, Systems Modeling and Control
;
Vision, Recognition and Reconstruction
Abstract:
Vision based techniques are by now well established means for contactless measuring applications, e.g. in industrial automation, optical inspection, medical imaging or robot navigation. However, the development and implementation of these applications on small-sized embedded systems is still challenging, because image processing algorithms require quite a lot CPU performance. Therefore, code optimization has to be considered in the development process. In this work we present our results on implementation and benchmarking of stereo disparity algorithms on an ARM/DSP embedded multicore platform. The algorithms are developed with the computer algebra framework MATLAB from Mathworks®, which allows to generate generic and processor-specific C/C++ code automatically. The analysis of the code generation process and the benchmark of the target performance are the main focus of this contribution.