Synthesis of Embedded Image Processing Applications from Simulink Models

In next generation video surveillance systems there is a trend towards embedded solutions. A significant amount of processing power is needed for such complex applications. Therefore, digital signal processors (DSP) are often used to provide the necessary computational capabilities. But resources in embedded DSP systems are typically very limited which imposes significant challenges for software development. Resource constraints must be met while facing increasing application complexity and pressing time-to-market demands. Recent advances in synthesis tools for SIMULINK suggest a feasible high-level approach to algorithm implementation for embedded DSP systems. The model-based visual development process of SIMULINK facilitates host-side simulation and validation, as well as synthesis of target specific code. Furthermore, legacy code written in MATLAB or ANSI C can be reused in custom blocks. However, the code generated for DSP platforms is often not very efficient. In this work the modeling and code generation capabilities within SIMULINK are evaluated with respect to image processing algorithms. A motion detection algorithm is synthesized from a corresponding model. The resulting code targeted at a Texas Instruments TMS320C6416 DSP is compared to a hand-optimized reference implementation for the same processor. Results show that an ad hoc approach to synthesize complex image processing algorithms hardly yields optimal code for DSPs. However, several methods are presented that can improve performance of synthesized code.