干涉多光谱图像压缩编码新技术

The New Image Compression Technology of the Interferential Spectral Image

提出一种基于运动估计的三维小波变换和非对称零树编码的干涉多光谱图像压缩方案.该方案利用大孔径静态干涉成像光谱仪推扫成像特点,对图像进行分区域运动估计的三维小波变换.采用了一种新型非对称零树编码方法,该方法可在较少的图像间建立较长的零树,增加了系数编码时由重要结点变为不重要结点并趋于零的概率,使编码的效果更好.该方案消耗内存少,延时小,有利于卫星上的图像压缩,有效地保护了图像的光谱特性.在8倍压缩下,满足干涉多光谱图像的质量要求.

A new compression scheme of interferential multispectral images is proposed based on the analyses of the large aperture static imaging spectrometry. There are two novelties in the scheme. One is that a three-dimension wavelet with special motion compensation is proposed. The other is that a new method of three-dimensional asymmetric zerotree coding （AT-3DSPIHT） is put forward. This AT3DSPIHT can make a longer tree in fewer images, which increases the probability of a coefficient value being zero when moving from root to leaves. Moreover, this method needs lower memory and shorter time delay than 3DSPIHT. Experimental results show that the scheme provides a better protection of the spectrum curve and a better performance on the compression.