基于纹理的雷达与多光谱遥感数据小波融合研究

众所周知,多光谱与雷达影像融合具有重要的意义,但雷达影像小尺度的纹理特征在先前的融合方法中却没有被考虑。为了更好地对多光谱与雷达影像进行融合,基于双正交小波变换,提出了一种小尺度纹理影像参与融合的三影像小波融合方法。该方法借鉴多通道滤波及基于亮度调节的平滑滤波(SFIM)融合的原理,首先提取多时相雷达影像的小尺度纹理数据;然后再将该纹理数据、单时相JERS-1SAR数据及TM多光谱数据进行小波融合。分析表明,该方法的融合结果较雷达与多光谱影像小波融合的结果不仅具有更丰富的光谱特征,而且由于继承了雷达影像丰富的小尺度纹理特征,因而具有更高的清晰度。实验证明,该方法可获得较好的融合结果,是一种切实有效的融合法。

Fusion Algorithm Research Based on Texture for SAR and Multi-spectral Images with Wavelet Transform

The fusion between multispectral and SAR(synthetic aperture radar)images has great significances.However,the small scale texture information of SAR images has not aroused scholars' attention in previous fusions.For better fusion result,a new data fusion technique has thus been developed based on biorthogonal wavelet transform with three types of images.They are:TM multi-spectral image,single temporal JERS-1 SAR image and the small scale texture image extracted from multi-temporal JERS-1 SAR images based on the principles of multi-channel filtering and SFIM(smoothing filter-based intensity modulation)fusion.Compared with the fusion method without texture image,the new one not only preserves spectral information well,but also inherits from SAR images rich small scale textural information that makes the result have higher definition.The wavelet fusion algorithm with texture image is of high flexibility and practicability.