Novel video coding algorithm based on 3D-binDCT

Abstract

In this paper we propose a three dimensional multiplierless discrete cosine transform (DCT) with lifting scheme called 3D-binDCT. Based on 3D-binDCT, a novel video coding algorithm without motion estimation/compensation is proposed. It uses the 3D-binDCT to exploit spatial or temporal redundancy. The computation of binDCT only needs shift and addition operations, thus the computational complexity is minimized. DC coefficient prediction, modified scan mode and arithmetic coding techniques are also adopted. Extensive simulation results show that the proposed coding scheme provides higher coding efficiency and improves visual quality, and it is easy to be realized by software and hardware.

References

[1] R. Westwater, and B. Furht, Kluwer Academic Publishers, Boston, 1997, 35–64.

[2] ChenW., HarrisonC., FralickS.. IEEE Trans. Communiccations, 1997, 25: 1004-1004

[3] LoefflerC., LightenbergA., mschytzG.. Proc. IEEE ICASSP, 1989, 2: 988-988

[4] FeigE., WinogradS.. IEEE Trans. Information Theory, 1992, 38: 1387-1387

[5] LiangJ., TranT. D.. IEEE Trans. Signal Processing, 2001, 49: 3032-3032

[6] BozinovieN., KonradJ.. SPIE symposium on Image and Video Commun. and Proc., 2003, 7: 1252-1252

[7] LeeM. C., ChanK. W., AdjerohD. A.. Journal of visual communication and image representation, 1997, 8: 405-405

[8] LeeH., KimY., RowbergA. H., RiskinE. A.. IEEE Trans. on Medical Imaging, 1999, 9: 478-478

[9] MarpeD., SchwarzH., WiegandT.. IEEE Trans. on Circuits and System for Video Tech., 2003, 13: 620-620