Memory bandwidth reduction for H.264 and HEVC encoders using lossless reference frame coding

Silveira, Dieison; Povala, Guilherme; Amaral, Livia; Agostini, Luciano; Porto, Marcelo

doi:10.1109/iscas.2014.6865711

Cited by 3 publications

(2 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…one pixel [7]. Compared with it, the methods based on 4x4 blocks [8,9] increase the memory access bandwidth remarkably (about 40%) when the align/  burst parameters are low. What's more, Flatness features of different block sizes are different, too.…”

Section: ) the Basic Access Unit Size Is Determined By Compression Unmentioning

confidence: 99%

See 1 more Smart Citation

Adaptive Granularity selection in Reference Picture Memory Compression

Ma¹,

Kang²

2015

Proceedings of the 2015 International Conference on Mechatronics, Electronic, Industrial and Control Engineering

View full text Add to dashboard Cite

Memory size occupation and memory access bandwidth are capital issues for high resolution video codec design because of the large scale of data, and the high power consumption especially for wireless terminals. An adaptive and random access-obeyed reference pictures memory compression (RPMC) scheme based on as small granularity as 2x2 blocks for 2-bit truncation is proposed in this paper to solve the problems. The impact of granularity in RPMC to memory access bandwidth during motion estimation and/or motion compensation is analyzed firstly. Then an adaptive RPMC scheme based on 2x2 block size is proposed, based on min-max scalar quantization (MMSQ). Finally, the results based on HM are provided which show that compared with the common used 4x4 block-based methods, little performance loss is introduced. At the same time, based on the adaptively merge scheme, the average memory access bandwidth is saved than the 4x4 block based method by up to 17%.

show abstract

Section: ) the Basic Access Unit Size Is Determined By Compression Unmentioning

confidence: 99%

“…[7] This kind of methods has been utilized by some hardware designers. [8], [9] However, this vlc table based method could not obey the principle of random access, which could increase complex computation and delay. Another technique of offset compensation is proposed to reduce the compression loss [10].…”

Section: Introductionmentioning

confidence: 99%