Non-Separable 3D Integer Wavelet Transform for Lossless Data Compression

摘要: This paper proposes a three-dimensional (3D) integer wavelet transform with reduced amount of rounding noise. Non-separable multi-dimensional lifting structures are introduced to decrease the total number steps. Since step contains operation, variance noise generated due operation inside is reduced. also investigates performance from various aspects such as 1) in frequency domain and those pixel domain, 2) rate distortion curve lossy coding mode entropy lossless mode, 3) computational time transforms, 4) feature comparison other methods. The proposed has merit that its output signal, apart noise, exactly same conventional separable structure which cascade 1D structure. Due this compatibility, it becomes possible utilize legacy previously designed transforms preferable properties regularity. Furthermore, expression signal values significantly because operations decreased by introducing non-separable includes memory accessing. It contributes increase system based on 3D transform. As result experiments, was observed method increases data compression input signals.