Sparse Scene Synthetic Aperture Radar Image Compression

Synthetic Aperture Radar makes huge number of data due to its characteristic of all-time and all-weather. General algorithms are designed to process optical images, which neglect the characteristic of SAR images. Inspired by Region of Interest (ROI) in JPEG2000, a compression algorithm based on the distribution of SAR images is proposed. After extraction of ROI using Otsu mathod, a quantization strategy is employed, through which ROI and background are retained to different extents according to user needs. Considering the sparsity of scene, pixels above threshold are mapped to a triple before entropy encoding for grayscales and locations. Then bit plane encoder (BPE) is employed considering the relatively small deviation of background. Experimental results on ocean-SAR images show that the Peak Signal Noise Ratio (PSNR) of proposed algorithm is 5~10dB higher than JPEG2000 at the same bit rate.Compressive sensing, as a vigorous method for signal processing, has not been wide studied in SAR data compression. In this article some research on ocean-SAR images using Orthogonal Matching Pursuit has been conducted. Results indicate that the sparsity of ocean-SAR images is suitable for compressive sensing.