SAR Raw Data Compression Based On Trellis Coded Quantization And Wavelet Transform

Synthetic Aperture Radar (SAR) is a kind of all-weather and day-and-night remote sensing tools, through which high resolution image can be obtained. SAR raw data is great amount and has wide dynamic range. The correlation between raw data samples is weak and the distribution of echo signals is approximately white Gaussian noise. These caused great difficulties to downlink and store SAR raw data. So the efficient methods for SAR raw data compression have become important tools in fulfilling the requirement of reducing data rate. In this article, some SAR raw data compression methods have been studied to improve the compression performance by using true airborne SAR raw data.In this paper, the statistics characteristic of SAR raw data is analyzed, which is the theoretic foundation of the raw data compression methods. And the traditional compression methods of Block Adaptive Quantization (BAQ),Block Adaptive Vector Quantization(BAVQ),FFT-BAQ and Discrete Wavelet Transform BAQ(DWT-BAQ) are reviewed and evaluated for the future use.Combining with traditional BAQ, Block Adaptive Trellis Coded Quantization (BATCQ) method which utilizes TCQ (Trellis Coded Quantization) in SAR raw data compression is proposed. At first, block adaptive quantization of raw data is implemented to reduce the data dynamic range. And then, based on the SAR raw data characteristic, TCQ is applied to the data after BAQ. This method utilizes convolutional encoding and signal space expanding to increase Euclidian distance between signals, and the gain of quantization is further improved. Simulation shows this method has advantages in SAR raw data compression in quantization and rate-distortion performance.Based on DWT-BAQ method, Lifting Wavelet Transform BATCQ (LWT-BATCQ) method is proposed in this paper. Combining the virtues of wavelet transform and TCQ, the method performs TCQ in wavelet domain and outperforms conventional method in terms of SNR. On the other hand, lifting wavelet replaces traditional wavelet to reduce time and memory consuming.Some subband wavelet coefficients are null in low bits rate compression method using wavelet transform. This causes spectral characteristics changing and spectrum losing. Due to this problem, LWT-BAQ/VQ method is proposed. Low dimensional vector quantization was assigned to low bit rate subband which increase few complexity. So the characteristics of the spectrum are maintained and the spectrum is kept. Data compression will not reduce the resolution of the radar images.Based on traditional wavelet transform methods, a Multiwavelet Transform SAR raw data compression method(MWT-BAQ) is presented for the first time in this paper. Multiwavelet technique overcomes the shortcomings of traditional wavelet and obtains better quantitative performance.