首页|Researchers Submit Patent Application, 'Systems And Methods For Hybrid Machine Learning And Dct-Based Video Compression', for Approval (USPTO 20230412796)
Researchers Submit Patent Application, 'Systems And Methods For Hybrid Machine Learning And Dct-Based Video Compression', for Approval (USPTO 20230412796)
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NETL
NSTL
News editors obtained the following quote from the background information supplied by the inventors:“Digital video compression formats have continued to evolve. In particular, the International TelecommunicationsUnion (ITU) and the International Standards Organization (ISO) have developed distinct videocodecs, including H.261, MPEG-1, H.263, MPEG-2, H.264, MPEG-4, and H.265/HEVC. All of the videocompression and decompression (codec) formats share two common themes: they use the discrete cosinetransform (DCT) and employ motion compensation to more effectively encode similarities that exist in thevideo, particularly between adjacent frames. Each iteration of MPEG video has added successively moretypes of search options to reduce the amount of residual information that needs to be included in the compressedbitstream. As video compression standards have continued to develop, the amount of computingpower to achieve the next standard has comparatively increased. For example, the latest HEVC/H.265standard uses approximately ten times the computing power compared to the last-generation H.264 codec.There is therefore an unmet need for alternative compression techniques.”
NETL
NSTL
