基于改进预测树的超光谱遥感图像无损压缩方法

该文在传统预测树方法的基础上提出一种改进方法，该方法定义一个幅度拉伸因子来表达相邻波段的局部灰度变化，通过比较局部上下文梯度来估算该幅度因子，并用它对当前的预测值进行修正。此外，还结合AVIRIS超光谱遥感图像的相关性特性提出一种谱间预测和空间预测相结合的综合预测无损压缩方案，在不同波段范围内采用可选的预测方式进行预测。在AVIRIS遥感图像数据集上的实验结果表明，该方案在计算复杂度较低的情况下，能够更好地消除冗余信息，具有良好的压缩性能。

The Lossless Compression Method for Hyperspectral Images Based on Optimized Prediction Tree

Prediction tree is a traditional and efficient method for lossless compression of hyperspectral image. In this paper an optimized method based on prediction tree is presented. To express the variation of local context of two neighboring bands, a partial extending factor is introduced to compensate the predicted value of current pixel so as to reduce the prediction error. Furthermore, a synthetical prediction based lossless compression scheme for AVIRIS hyperspectral images is proposed. Experimental results demonstrate that the proposed method works efficiently on AVIRIS images with low complexity and limited memory.