基于重要系数提升的遥感图像压缩

提出了一种基于重要系数确定及系数提升幅度计算的遥感图像压缩方法。根据图像小波域系数的熵估计值确定正常编码情况下图像编码所截取到的比特平面,然后对重要的系数进行比特平面提升,这样系统编码时就能够保留更多的小目标信息。本算法与感兴趣区域编码方法不同,重要系数的定义及其提升幅度大小是根据图像内容和输出码率确定的,可以认为本算法是任意区域的感兴趣区域编码的扩展,但更为灵活。仿真结果表明,本算法可以有效保留遥感图像中小目标的信息, 系统重建图像质量略有下降,克服了现有图像压缩算法一般以均方误差最小为准则,在低码率时会失去小目标信息细节的局限。

A remote-sensing image coding algorithm based on significant coefficients lifting

MSE(Minimum Mean Square Error) is widely used as the criteria in image coding algorithm for providing much detail and important information of small target.In order to remain more detail information,an algorithm was proposed for determining the significant coefficients and the lifting value.The final coding bit-plane can be estimated by the entropy estimation of the DWT coefficients,and also the significant coefficients can be defined and the lifting value can be determined.The difference between proposed algorithm and ROI(Region-of-Interest) is that the definition of significant coefficients and the lifting value can be obtained from the coding image and the given coding rate so that the algorithm can be regard as the generality of ROI coding algorithm,but more flexible.The experimental results show that the small targets in remote-sensing images can be saved validly,while the detail information in reconstructed image are lost a little.