Image Processing
After the completion and validation of a software prototype real-time image processing solutions for cost-optimised Embedded Systems pose additional challenges to the developers:
- Optimization of memory usage and processor load: Where can images be replaced by line-oriented algorithms, up to which stage of processing is access needed to the original image data?
- Replacement of floating point by integer arithmetic: What scaling and word lengths maintain the accuracy of the results?
- Partitioning algorithms for parallel processing in multicore architectures: Where should the system be cut to balance the load between the cores, how much delay is created by communication between the cores?
- Proof of compliance with the real-time requirements: Can the target system calculate its answers in the image timing regardless of the content of the video stream?
- Evidence of functional equivalence: Does the target system return the same answers as the prototype given the same video data?
Solving these problems requires close collaboration between process developers, porting providers and hardware developers. With its experiences of various projects for the processing of real scenes FORTecH Software has developed a methodology for the safe, efficient transfer of prototype solutions into production-ready products. Automatic testing ensures continuous verification of the target code.