基于二代小波变换和PCNN的高光谱图像融合算法

采用信息融合技术可以降低高光谱遥感图像的分析难度。本文提出一种基于二代小波变换和脉冲耦合神经网络(PCNN)的融合算法。在利用自适应子空间分解技术将高光谱图像的数据空间划分为数个子空间后,对各子空间内的每一波段图像进行二代提升小波分解。对低频系数部分进行方差加权融合的同时利用PCNN的脉冲同步和全局耦合特性对高频系数部分进行选取,最后用二代小波逆变换得到各子空间的融合图像.其仿真实验结果显示:所提算法有效降低了高光谱图像维数,很好保留了原图像的信息,效果优于单一的一代小波和二代小波融合算法。

Fusion algorithm of hyperspectral image based on the second generation wavelet transform and PCNN

Fusion technology was an effective way to reduce the difficulty of imagert processing. Here, a new fusion algorithm based on the second generation wavelet transform and pulse-coupled neural networks (PCNN) was proposed. After divided the whole data space into several subspaces by adaptive subspace decomposition technology, each band image was decomposed into low frequency coefficient and high frequency coefficient by the second generation wavelet transform. Then the low frequency coefficient was weighted fused according to the standard deviation of each image, and in the high parts, use the global coupling and pulse synchronization characteristics of PCNN to choose high frequency coefficients. The result of the fusion experiment based on the real hyperspectral data showed :the new algorithm can get satisfied fusion effects,and the result is better than the fusion algorithms that use the first generation wavelet or the second generation wavelet singly.