CPUs, GPUs, FPGAs: How to choose the best method for your machine vision application

摘要

As these examples show no one-size-fits-all platform for machine vision applications exists. Many variables influence whether a CPU, GPU, or FPGA—or some combination of the three—should be selected. Smart cameras may employ CPUs, DSPs, or a combination of CPU and FPGA. A broad range of options regarding power consumption and processing speed may exist within a single platform. Best performance determines the correct solution.