基于快速层次交替最小二乘非负张量Tucker分解的干涉高光谱图像光谱信息压缩方法

提出一种基于快速层次交替最小二乘非负张量Tucker分解的高光谱图像光谱信息压缩算法。首先，将干涉高光谱图像光程差方向的三维信息采用三维光程差方向提升小波变换(3D OPT-LDWT)进行分解，将三维小波子带系数看作三阶非负张量，采用快速层次交替最小二乘非负张量Tucker分解(FHALS-NTD)算法对进行分解，得到核心张量和模式矩阵。对每个模式矩阵进行量化，对核心张量采用比特平面重要系数编码算法进行编码，得出最终的压缩码流。结果表明，此压缩算法可以稳定可靠地工作。与传统压缩算法比较，平均信噪比提高了1.23 dB。有效的提高了干涉高光谱图像压缩性能。

Compression of Interference Hyperspectral Image Based on FHALS-NTD

A hyperspectral interference image compression algorithm based on fast hierarchical alternating least squares nonnegative tensor Tucker decomposition (FHALS-NTD) is proposed. Firstly, the interference hyperspectral image is decomposed by 3-D OPD lifting-based discrete wavelet transform (3D OPT-LDWT) in the OPD direction. Then, the 3D DWT sub-bands decomposed are used as a three order nonnegative tensor, which is decomposed by the proposed FHALS-NTD algorithm to obtain 8 core tensors and 24 unknown component matrices. Finally, to obtain the final compressed bit-stream, each unknown component matrices element is quantized, and each core tensor is encoded by the proposed bit-plane coding of significant coefficients. The experimental results showed that the proposed compression algorithm could be used for reliable and stable work and has good compressive property. In the compression ratio range from 32∶1 to 4∶1, the average peak signal to noise ratio of proposed compression algorithm is higher than 40 dB. Compared with traditional approaches, the proposed method could improve the average PSNR by 1.23 dB. This effectively improves the compression performance of hyperspectral interference image.