A comparison of hardware/software techniques in the speedup of color image processing algorithms

The speedup of color image processing algorithms remains of high interest as image resolutions continue to increase. Due to the intrinsic massive data parallelism, the techniques employed use multi-core processor systems or custom hardware. In the former the task is split in multiple threads at the software level, allowing as many data sets to be processed, as there are threads available. In the latter the concurrency is limited only by the size of the available hardware resources. Both techniques have their advantages and disadvantages, and have been demonstrated in isolation. However, the advent of heterogeneous computing platforms, i.e. the combination of a processor based system (PS) and programmable logic based system (PL), enables a comparison of these techniques on the same kind of platform. To this end, the work presented in this paper compares and analysis three color image processing implementations based on: the use of SIMD instructions, multithreading, and custom hardware.