DocumentCode :
3055068
Title :
Low bit-rate video coding with trellis source codes
Author :
Mello, Magatho A. ; Fischer, Thomas R.
Author_Institution :
Dept. of Electr. Eng. & Comput. Sci., Washington State Univ., Pullman, WA, USA
fYear :
1998
fDate :
30 Mar-1 Apr 1998
Firstpage :
562
Abstract :
Summary form only given. A novel trellis coded quantisation (TCQ) encoding method is introduced that uses a hard thresholding operation to partition the zig-zag scanned coefficients into two classes: significant, |x|⩾T, and insignificant, |x|<T, where T is a threshold. Only significant coefficients are TCQ encoded, the remainder being scaler quantized to zero. Aside from improved rate-distortion performance, the thresholding reduces the algorithmic complexity, since it is only the (relatively few) significant DCT coefficients that need be trellis encoded. However, to achieve a significant portion of the available granular gain, it is necessary to TCQ encode a large number of coefficients. Thus, an entire GOB is jointly encoded. An arithmetic code is used to encode the sequence of zero-runs and non-zero TCQ codeword indices. Comparing the performance of the new method to Telenor´s H.263 algorithm for intra- and inter-frame coding, at medium to large encoding rates, the new method provides improvements of up to about 1 dB. At small encoding rates the combination of diminishing granular gain available over scalar quantization, the relatively few significant DCT coefficients in a typical group of blocks, and the overhead of truncating the arithmetic coding after each frame, causes the new method to perform no better than, and sometimes worse than, the H.263 algorithm, for encoding in the inter-frame mode
Keywords :
arithmetic codes; discrete cosine transforms; rate distortion theory; source coding; transform coding; trellis codes; video coding; GOB; TCQ; algorithmic complexity; arithmetic code; codeword indice; granular gain; inter-frame coding; intra-frame coding; low bit-rate video coding; performances; rate-distortion performance; scalar quantization; significant coefficients; thresholding operation; trellis coded quantisation encoding; trellis source codes; truncation; zero-runs; zig-zag scanned coefficients; Arithmetic; Circuits and systems; Computer science; Discrete cosine transforms; Encoding; Partitioning algorithms; Performance gain; Quantization; Rate-distortion; Video coding;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Data Compression Conference, 1998. DCC '98. Proceedings
Conference_Location :
Snowbird, UT
ISSN :
1068-0314
Print_ISBN :
0-8186-8406-2
Type :
conf
DOI :
10.1109/DCC.1998.672304
Filename :
672304
Link To Document :
بازگشت