基于K-均值聚类和传统递归最小二乘法的高光谱图像无损压缩

针对基于预测的高光谱图像无损压缩算法压缩比低的问题，该文将聚类算法与高光谱图像预测压缩算法相结合，提出一种基于K-均值聚类和传统递归最小二乘法的高光谱图像无损压缩算法。首先，对高光谱图像按光谱矢量进行K-均值聚类以提升同类光谱矢量间的相似度。然后，对每一聚类群分别使用传统递归最小二乘法进行预测，消除高光谱图像的空间冗余和谱间冗余。最后，对预测误差图像进行算术编码，完成高光谱图像压缩过程。对AVIRIS 2006高光谱数据进行仿真实验，所提算法对16位校正图像、16位未校正图像和12位未校正图像分别取得了4.63倍，2.82倍和4.77倍的压缩比，优于同类型已报道的各种算法。

Lossless Compression of Hyperspectral Images Using K-means Clustering and Conventional Recursive Least-squares Predictor

To improve the compression ratio of lossless compression scheme based on prediction, a lossless compression scheme for hyperspectral images using K-means Clustering method and Conventional Recursive Least-Squares (C-CRLS) predictor is presented in this paper. The proposed scheme first clusters the spectral data into clusters according to their spectra using the famous K-means clustering method. Then, the proposed scheme calculates the preliminary estimates to form the input vector of the conventional recursive least-squares predictor. Finally, after prediction, the prediction residuals are sent to the arithmetic coder. Experiments on the Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) 2006 hyperspectral images show that the proposed scheme yields an average compression ratio of 4.63, 2.82, and 4.77 on the 16-bit calibrated images, the 16-bit uncalibrated images, and the 12-bit uncalibrated images, respectively. Experimental results demonstrate that the proposed scheme outperforms other current state-of-the-art schemes for hyperspectral images that have been previously reported.