S.B. Borbély, J. Kogler, and W. Kubinger (Austria)
Computer Vision, Embedded and Real Time Systems, Modeling and Simulation Languages, Automatic Code Generation.
A growing demand for Embedded Computer Vision Ap plications exists to tackle problems which can hardly be solved by traditional sensor systems. However, due to the high computational complexity of most computer vision algorithms, realization of such applications on embedded systems is not a straight forward task. This paper describes a proposed design process on model-based design for embedded computer vision appli cations. The main goal is to accelerate development and to improve the quality of the resulting applications by sys tematic design, development and test. To demonstrate and describe the design flow, case studies based on stereo image processing algorithms have been chosen as example. The whole design flow is simulated under the MAT LAB Simulink environment. Target specific code genera tion is performed after Model In the Loop (MIL) and Soft ware In the Loop (SIL) simulation. The targets (two Texas Instruments DSP Starter Kit (DSK) boards) are flashed with the generated code. The results are compared and evaluated based on a Processor In the Loop (PIL) and a Hardware in the Loop (HIL) system run. For instrumenta tion of the generated code, the used CPU cycles of the core part of the algorithm are counted as a measure for the mean execution time.
Important Links:
Go Back
