Abstract
Title : A Best Wavelet Packet Basis for Joint Image Deblurring-denoising and Compression
Authors : Pierre Dhérété, Sylvain Durand, Jacques Froment and Bernard Rougé
We propose a unique mathematical framework to deblur, denoise and compress
natural images. Images are decomposed in a wavelet packet basis adapted
both to the deblurring filter and to the denoising process. Effective denoising
is performed by thresholding small wavelet packet coefficients while deblurring is
obtained by multiplying the coefficients with a deconvolution kernel.
This representation is compressed by quantizing the remaining coefficients and by
coding the values using a context-based entropy coder.
We present examples of such treatments on a satellite image chain.
The results show a significant improvement compared to separate treatments
with up-to-date compression approach.