[Sls-mhdc] Entropy Coder Simplification (MHDC-A-1610-5(a)(3))

[Kiely, Aaron B (332B)]

Dear Working Group,

In response to action item MHDC-A-1610-5(a)(3), I have uploaded to the CWE the following document:
	CWE Private / 123.1-B / EntropyCoder-Simplification.pdf

The document describes Ian’s idea of a way to simplify the FLEX Interleaved Entropy Coder (IEC), completely eliminating the queue(!), and presents some performance results comparing compression effectiveness of the original and simplified IECs.

We have not implemented this in hardware, but I’ve discussed it with the person who did our FPGA implementation of the original FLEX entropy coder, and his impression is that this simplified version should be much faster.

My opinion is that this simplification addresses the desire of the WG to have a faster entropy coder, and the coding effectiveness of this simplified version is acceptable. So my recommendation is that we adopt this.

Please let me know your opinion.

I think a key question is whether the simplified entropy coder would prevent (or make it very difficult for) us to implement any of the optional features under consideration (relative error preservation, HVS quantization, rate control). I hope that those of you who are experts on these features can comment on whether the simplified IEC raises concerns.

Can reach a consensus by email in the next week or so?

Regards,
Aaron