Abstract:
Stereo vision processing is relatively resource intensive in order to produce a reasonable good 3D result. This processing requires powerful CPUs and GPUs to render a collection of original images using complicated algorithms. Only powerful servers are capable of completing this task in short period of time. Original images and processed results need to be transferred between client and servers. A slow network connection may counteract the advantage of server processing speed. Recent evolution of microprocessor technology gives an opportunity to embed powerful CPUs and GPUs into portable devices. So a powerful smart phone processor is capable of completing the stereo vision processing. Although this processing still needs a longer time to execute on portable devices than on servers, original images and results do not necessarily needed to transfer through network.
