Narrowing the performance gap between orthogonal and biorthogonal wavelets

Recent techniques have suggested the possibility of implementing symmetric extension with orthogonal wavelets enabling us to address some unresolved image compression questions (V. Silva et al., 1999). This paper demonstrates that linear phase filters are critical to compression performance - an issue not previously recognized. Moreover, symmetric extension outperforms periodic extension for both biorthogonal and orthogonal wavelets. Biorthogonal and orthogonal wavelets with similar filter properties typically generate similar compression performance when both employ symmetric extension. However, biorthogonal wavelets show a slight performance advantage for low frequency images; this advantage is significantly smaller than previously published results and is explained in terms of wavelet properties not previously considered.