NLE System Design Using Mathematically Lossless Motion-JPEG

Most professional video applications, including nonlinear editing systems (NLE), digital disk recorders (DDR) and media servers systems, employ the ITU-601 video standard. The uncompressed, digitized 601 video signal contains 20 Mbytes of data per second of video. Storing or retrieving that much data to/from a hard disk in realtime demands a very high-performance system with high bandwidth characteristics throughout the entire system and massive storage devices, leading to a very expensive implementation. Until now, only high-end post-production houses have been able to afford such systems. — In order to reduce system cost, Motion-JPEG compression has been used in many system implementations to reduce data rate and storage requirements for digital video streams. However, the compression process introduces loss of information in the video signal. Video quality is inversely proportional to the compression factor—a higher compression factor leads to lower video quality, with more compression artifacts. Until now, users have had to make the tradeoff of either paying the very high price of uncompressed video systems or accepting the video-quality compromise inherent in affordable systems. — Early in 1996, Matrox introduced a digital video platform called DigiSuite that offers the quality of uncompressed video at a reasonable cost. Using mathematically lossless encoding, DigiSuite performs data reduction on the video image using the entropy encoding part of the M-JPEG standard. Entropy encoding introduces no artifacts or loss of information—the process is MATHEMATICALLY lossless, i.e. the digital binary code of the decoded image is identical to the source image. Users benefit from the data rate and storage space reduction of M-JPEG without the tradeoff of introducing artifacts into the video signal. — This paper describes the theory behind lossless M-JPEG encoding and goes on to examine the practical implementation issues that must be taken into account in system design. All the individual components such as the M-JPEG chip set, the disk controller and the storage technology are examined. As well, a system-level hardware and software architecture that is capable of performing data stream management and data buffering at the necessary speed is described. In addition, a test suite is described that can be used to prove that the compression method is, indeed, lossless.

Published

1997-02

Content type

Original Research

DOI

10.5594/M00200

ISBN

978-1-61482-926-3