UHD Panoramic Video Coding for Multi-Camera and Multi-Processor Acquisition Systems
Carreira, J. F. M.
UHD Panoramic Video Coding for Multi-Camera and Multi-Processor Acquisition Systems, Proc The Thirteenth International Conference on Mobile Ubiquitous Computing, Systems, Services and Technologies UBICOMM, Porto, Portugal, Vol. , pp. - , September, 2019.
Digital Object Identifier:
This paper addresses the problem of encoding Ultra-High Definition (UHD) panoramic video in multi-camera and multi-processor systems, where a wide Field of View (FoV) is captured by multiple cameras, each one corresponding to a small FoV. The UHD panoramic images are then formed by stitching high-definition images captured by the individual cameras, covering a wide FoV around the acquisition viewpoint. A simulation study is carried out to evaluate the rate-distortion-complexity performance of multi-encoder systems using independent processors for encoding sub-images with small FoV, as part of a wider panoramic UHD resolution. This study uses an image partitioning scheme to distribute the UHD images with wide FoV over various processors and evaluates the rate-distortion-complexity performance of HEVC encoding in comparison with classic encoding of a single frame per processor. The results show that the rate-distortion performance of multi-processor systems is quite similar to single processor ones, which allows to distribute the huge computational requirements of HEVC encoding across several low-cost processors.