Chapter 1: Invited Addresses and Tutorials on Signals, Coding,
  Systems and Intelligent Techniques

Title: Context based quantization using Bayesian inference

Author(s): Dušan Gleich, Mihai Datcu

Address: Remote Sensing Institute, German Aerospace Center, Oberpfaffenhofen, 82234 Weßling | Remote Sensing Institute, German Aerospace Center, Oberpfaffenhofen, 82234 Weßling

Reference: 12th International Workshop on Systems, Signals and Image Processing pp. 459 - 462

Abstract/Summary: This paper presents a transformed-based compression of high resolution Synthetic Aperture Radar (SAR) images using wavelet transformation. SAR images are corrupted by multiplicative noise called speckle. To achieve higher compression ratio, this noise is firstly removed and denoised image is compressed. To remove speckle from SAR images a Bayesian filter in wavelet domain is developed. Lossy quantization is implemented using Bayesian theory in order to predict observed wavelet coefficient using inter and intra-scale dependencies in wavelet domain. The predicted indices are encoded. The signal-to noise ratio of reconstructed images and achieved compression ratios are compared with the state-of-the-art compression methods applied to the SAR images with resolution of one meter.

Order a copy of this article