基于平稳小波变换的图像融合算法

Image Fusion Algorithm Based on Stationary Wavelet Transform

针对IHS(intensity,hue.saturation)融合算法的光谱畸变和正交小波变换融合算法的方块效应问题,提出了一种基于平稳小波变换的图像融合算法,并且探讨了在图像处理中应用灰色系统理论的可行性和有效性。该算法结合IHS变换,对于图像平稳小波分解的高频分量利用灰色关联分析准确、有效地检测出有用的边缘信息,并在确定融合权值时,给予不同的重视程度。实验表明,本文算法能够克服正交小波变换在图像融合中出现的方块效应问题,在保持光谱信息方面具有显著的优越性,并且能够有效提高融合图像的空间分辨能力,又由于新算法引入灰色关联分析进行像素分类,具有良好的抗噪性和可控边缘信息量等优点,使算法灵活、有效,进一步改善了融合图像的质量。

Fusion algorithms have the following disadvantages： （1） much loss of spectrum information by IHS（intensity, hue, saturation） transform; （2） mosaic occurred on the fusion image. So a better image fusion algorithm is proposed based on stationary wavelet transform to fuse a TM multi-spectral and low-resolution image with high-resolution SAR image. This paper discusses a novel application of grey relation analysis in image processing. Firstly, the grey relational analysis is used to classify pixels on the high-frequency-image decomposed by stationary wavelet transform into non-edge pixels and edge pixels. Then, by stationary wavelet transform a SAR image with I component obtained from IHS transform of TM image with different weighting strategies is fused, the weight for edge pixels is higher than that for non- edge pixels. Finally, the inverse stationary wavelet transform is carried out. The experimental result indicates that the new method is flexible and can resist noises better, thus enhancing spatial detail information of the fusion image efficiently and spectral information of the original TM image much better.