Affiliations: [a] Institute of Applied Mathematics, Hebei Academy of Sciences, Shijiazhuang, China
| [b] Hebei Authentication Technology Engineering Research Center, Shijiazhuang, China
| [c] Department of Mathematics and Physics, Shijiazhuang Tiedao University, Shijiazhuang, China
| [d] The center for SCDM, NIT, Zhejiang University, Ningbo, China
Correspondence:
[*]
Corresponding author. Ruiqin Fan, Department of Mathematics and Physics, Shijiazhuang Tiedao University, Shijiazhuang, 050043, China. E-mail: fanruiqin@126.com.
Abstract: Wavelet synopses with maximum error bound is an effective quality-guaranteed compression method that restricts the approximation error of each data does not exceed a given error bound. In this paper, we focus on the study of constructing efficient two-dimensional wavelet synopses with maximum error bound. First, we propose a linear-time two-dimensional F-shift algorithm (TDFS), then present a general parallel framework for two-dimensional data array and generate a parallel two-dimensional F-shift algorithm (PTDFS). We have proven that the size of a synopsis constructed from PTDFS is always no larger than that of the existing methods and can reduce up to 66.7% at most. The experimental results indicate that the synopsis sizes can be reduced from 40% to 60% in most situations. Moreover, PTDFS can not only improve the quality of reconstruction image, but also reduce the running time.
Keywords: Wavelet synopses, parallel, error bound, data compression
