Morphological wavelets for 3D volume image decorrelation

Abstract

The performances of morphological wavelets for 3D volume image decorrelation in lossless image compression are presented. Nonlinear morphological wavelets used in non-redundant multiresolution signal decomposition schemes with perfect reconstruction are very fast for implementation as only integer arithmetic is used. The lifting schemes implementing morphological wavelet transforms avoid quantizers which is attractive property for lossless data compression. The performances of both separable and non-separable 3D morphological wavelet transforms using lifting scheme is evaluated. The computation time of 3D decomposition implemented by morphological wavelets is 3 times shorter than the decomposition time with 5/3 wavelet filters used in JPEG2000 standard with similar decorrelation efficiency for lossless compression.