Variable block size multistage VQ for video coding

In this paper, a video coding algorithm based on multistage vector quantization (MVQ) and rate distortion functions is proposed. Each frame is divided into 16/spl times/16 blocks. Each block is encoded using variable block size MVQ and/or block-based motion compensation. For the MVQ coding, each block is partitioned into blocks (or vectors) of size 256, 128, 64, 32, 16, and 8 pixels. The rate-distortion measurements associated with the different block sizes are used to select the partitioning of a particular block. The bit rate allocated to each frame depends upon a buffer fullness function as well as rate-distortion information for the frame. The proposed algorithm has low computational complexity for the decoder.

[1]  Jorma Rissanen,et al.  Applications of universal context modeling to lossless compression of gray-scale images , 1995, Conference Record of The Twenty-Ninth Asilomar Conference on Signals, Systems and Computers.

[2]  Guillermo Sapiro,et al.  LOCO-I: a low complexity, context-based, lossless image compression algorithm , 1996, Proceedings of Data Compression Conference - DCC '96.

[3]  Gary J. Sullivan,et al.  Efficient quadtree coding of images and video , 1994, IEEE Trans. Image Process..

[4]  J. Rissanen,et al.  Applications of universal context modeling to lossless compression of gray-scale images , 1995 .

[5]  Yair Shoham,et al.  Efficient bit allocation for an arbitrary set of quantizers [speech coding] , 1988, IEEE Trans. Acoust. Speech Signal Process..

[6]  Richard W. Harris,et al.  Comparison between adaptive search and bit allocation algorithms for image compression using vector quantization , 1995, IEEE Trans. Image Process..

[7]  Syed A. Rizvi,et al.  Advances in residual vector quantization: a review , 1996, IEEE Trans. Image Process..