基于小波包变换的分频图像融合方法

A Fusion Algorithm of Image Based on Wavelet Package and Different Frequencies Processing

利用多源遥感数据有利于以多种数据集互补的方式提供单一数据集无法提供的信息,图像融合则是多源信息集成的关键之一。本文提出了基于小波包变换的分频图像融合流程,即对于用小波包分解得到的低频和高频部分,采用不同的融合策略。采用熵、联合熵、偏差指数和边缘指数等作为定量指标,对这种流程的融合结果进行了质量评价。通过ETM+多光谱和全色图像融合实验表明这种方法在光谱信息保持和纹理信息增强上相较与传统融合方法有明显优势。

The use of multi-source remote sensing images is beneficial as multiple datasets can complement each other by providing an alternative or complementary means when the information in question is not reliably found from one dataset. Image fusion is one of the key techniques to integrate different types of sources together. An approach for image fusion based on wavelet package and different frequency processing is proposed in this paper. Entropy, cross entropy, difference index and edge index are employed to estimate the quality of the result image generated by fusing ETM ＋ （ Enhanced Thematic Mapper） multi-spectrum image and panchromatic image. Experimental result shows this approach is good at spectral reservation and texture enhancement.