基于局部上下文的超光谱图像压缩方法

Local Context Application to Hyper-spectral Image Compression

根据超光谱遥感图像有很强谱间相关性成像特点,及局部上下文三维预测方法的思想,设计三种适合超光谱图像的预测方法,并进行比较,选择其中的谱间LOCO-I预测方法。该方法是将前一波段像素值作为预测像素的基准值,根据谱间像素变化规律来选择修正梯度项。实验表明该预测方法具有很好的压缩效果。

According to hyper-spectral images having strong spectral correlation, and idea of local context 3D prediction model, an application of multi-layered decomposition compression is proposed. Three prediction models are designed, we choose spectral LOCO-I prediction model, the former band pixel value as base one, then choice the modified value by changing rules, The experimental results shown that the algorithm was efficient.