基于易操纵金字塔的多传感器图像融合

针对目前传统方法在图像未完全配准时融合效果差的问题,提出了一种基于易操纵金字塔的多传感器图像融合方法。首先,对多光谱图像进行易操纵金字塔分解;然后,恰当地合并分解得到子带图像系列来构造融合图像对应的易操纵金字塔,并通过逆变换重构融合后图像。最后利用熵和空间频率对该方法的融合性能进行了评估分析,并与基于拉普拉斯变换和小波变换的图像融合方法进行了比较。实验结果表明,该方法综合性能优于基于拉普拉斯变换和基于小波变换等传统图像融合方法,图像未完全配准情况下也能获得好的融合效果。

Multi-sensor Image Fusion Based on Steerable Pyramid

This paper introduces a novel steerable-pyramids-based multi-sensor image fusion method,which can be used to get better fusion result when the multi-sensor images are not correctly registered.First,the authors decompose the multi-sensor images using steerable pyramid method.Then,a new steerable pyramid is built by appropriately combin-ing the sub-bands.Finally,the output image is obtained by using inverse transform.They use entropy and spatial frequen-cy to evaluate the fusion result of their method and perform a comparison with the laplacian based and the wavelet based image fusion methods.Experiments demonstrate that their method can obtain a promising fusion result and it is superior to the laplacian based and the wavelet based image fusion methods.