A Novel ANS Coding with Low Computational Complexity

Asymmetric numeral systems (ANS) is a recent family of entropy coders, which provides the compression ratio of arithmetic coding with a processing speed similar to that of Huffman coding. It has been popular and incorporated into many compressors, including Facebook Zstandard, Apple LZFSE and Google Draco 3D. However, the encoding and decoding procedures of the ANS coding involve a large number of multiplication, division and modulo operations, resulting in a high computational demand. In this paper, we propose a novel ANS coding, which combines the truncated binary coding and the ANS coding to reduce computational complexity. The simulation results show that with an average loss of 1.1% in compression quality, the proposed method can fully avoid the usage of multiplication and division operations.