海洋合成孔径雷达图像的一种高效压缩方法

An efficient compression algorithm for ocean-SAR image

摘要:

针对通用压缩算法未利用合成孔径雷达(SAR)图像特征的不足，提出一种基于概率分布的自适应海洋SAR图像压缩算法。利用海洋SAR图像的概率分布，根据目标的分布设计量化方案，使目标和背景得到不同程度的保留。利用场景的稀疏性，将阈值以上的像素映射到三元组，对其灰度和位置信息分别熵编码；利用剩余背景层灰度偏差较小的特点作位平面编码。实验结果表明，该算法能有效地压缩图像，同码率下峰值信噪比(PSNR)较JPEG2000高5 dB~10 dB。本文算法复杂度低，对比度保持好，适用于针对不同需求的海面舰船SAR图像压缩。

Abstract:

Considering the defection of general compression algorithms which neglect the characteristics of Synthetic Aperture Radar(SAR) images, an adaptive compression algorithm based on the distribution of ocean-SAR images was proposed. According to the distribution, a quantization strategy was employed, through which target and background were retained respectively. Using the sparsity of the scene, pixels above the threshold were mapped to a triple before entropy encoding for both grayscales and locations. Then Bit Plane Encoder(BPE) was employed considering the relatively small deviation of the remaining background. Experimental results showed that the Peak Signal Noise Ratio(PSNR) of proposed algorithm was 5 dB-10 dB higher than JPEG2000 at the same bit rates. The proposed algorithm, with better contrast and lower computational complexity, is well suited to various requirements of ocean-SAR image compression.