Abstract :
In this paper we describe a practical approach to processor selection for embedded computer vision applications. This approach is based on expected production volumes and other requirements. We then present several popular lowlevel computer vision techniques that have been algorithmically optimized for implementation on an emerging class of embedded architectures, media processors.
Keywords :
"Computer vision","Parallel processing","VLIW","Costs","Hardware","Application software","Production","Computer architecture","Signal processing algorithms","Image processing"
