Wavelet packet and TCQ coding for sar raw data compression

Synthetic aperture radar (SAR) is an imaging system which provides high resolution images of earth surface. The high resolution in the range direction is achieved by using large bandwidth signals and that in the azimuth direction is achieved by synthesizing a large aperture antenna using platform motion. The unique data collection geometry of SAR system requires that huge amounts of raw data be processed before obtaining a viewable image. Therefore, performing some form of compression on SAR raw data provides an attractive option for SAR systems. In this paper, we present a transform coding approach for SAR raw data compression. Due to presence of large dynamic range of frequency content in SAR raw data we propose the usage of wavelet packet transform. Furthermore, the transform coefficients are quantized using universal trellis coded quantizers. The quantization is performed independently on each subband in contrary to the current image coding algorithms like JPEG2000, considering the nature of SAR raw data where dependencies across scales are not evident. An adaptive rate allocation scheme is used to efficiently allocate the fixed rate resources among different subbands. Experimental results of the proposed method provides significant improvement in SNR results over standard block adaptive quantization (BAQ) and JPEG2000 techniques.