EuroVis'11 Proceedings of the 13th Eurographics / IEEE - VGTC conference on Visualization
| Hi-index | 0.00 |
Image compression techniques are routinely appliedto conserve storage space and minimize bandwidthutilization in various video and communicationapplications. Wavelet transform is an efficient approachto reduce spatial redundancies without the annoyingblocking artifacts at low bit rates. The underlying signalprocessing used in wavelet transform is the convolutionbetween the decimated input signal and the waveletfilters. Since image signals are not continuous at theboundaries, problems of coefficient expansion andboundary distortion are faced in the implementation ofthe filtering on the finite length signal. Circularconvolution instead of linear convolution can eliminatecoefficient expansion but introduce boundary artifacts,especially when more levels of decompositions areinvolved to obtain scalable images. Symmetric-extendedwavelet transform (SWT) introduced in this paper cangreatly reduce boundary artifacts with improvedperformances in scalable image compression incomparison with the periodic-extended wavelet transformand the JPEG compression.