The LiftLT: fast-lapped transforms via lifting steps

摘要: This paper introduces a class of multiband linear phase-lapped biorthogonal transforms with fast, VLSI-friendly implementations via lifting steps called the LiftLT. The transform is based on lattice structure that robustly enforces both phase and perfect reconstruction properties. coefficients are parameterized as series steps, providing efficient, in-place computation coefficients. new designed for applications in image video coding. Compared to popular 8/spl times/8 DCT, times/16 LiftLT only requires one more multiplication, 22 additions, six shifting operations. However, coding examples show far superior DCT objective subjective performance. Thanks properly overlapping basis functions, can completely eliminate annoying blocking artifacts. In fact, novel LT's performance consistently surpasses much complex 9/7-tap wavelet floating-point More importantly, transform's block-based nature facilitates one-pass sequential block coding, region-of-interest coding/decoding, parallel processing.