提出一种基于“谱间预测+JPEG-LS”框架的星上未配准多光谱图像压缩方法。首先结合压缩方案,提出基于谱段重排的整像素谱段配准预处理算法,将谱段重排与谱段配准有机结合,显著提高了谱间相关性和压缩性能;其次提出基于二次迭代调整的码率控制算法,保证输出码率在接近目标码率的前提下严格低于目标码率,更好地满足固定链路带宽的要求。实验表明该压缩方法兼容无损和有损压缩,压缩性能优于其他方法,且算法实现简单。
Based on the framework of inter-band linear prediction and JPEG-LS, a compression scheme for unregistered satellite multispectral images is proposed. First of all, an integer-pixel band to band registration pre-processing method based on band reordering is presented according to the compression scheme to improve the poor spectral correlation and the compression performance of the multispectral images. Secondly, a rate control method based on twice-iterated adjustment is proposed to guarantee the output rate below the target rate strictly while approaching the target rate, so as to meet the transmission requirement better. Experimental results show that our proposed scheme is compatible with lossless and lossy compression, and outperforms other methods with simple implementation.