基于多进制小波变换及多维纹理特征的遥感影像融合方法

摘要

探讨了基于多进制小波变换与多维纹理特征融合相结合的遥感影像融合方法.在融合过程中,首先对高分辨率全色影像和多光谱影像进行多进制小波分解,再联合提取局部方差、局部梯度、局部能量和局部信息熵4维纹理特征,将高分辨率影像的高频分量分别与多光谱影像的高频分量以多维纹理特征进行多判据联合方法融合,形成新的高频分量,然后与多光谱影像的低频分量进行多进制小波逆变换,最后经RGB合成为彩色影像.试验选取淮南矿区SPOT 10 m与TM 30 m空间分辨率影像,从目视判读(定性评价)、地物光谱曲线分析、定量评价指标三方面对融合方法进行了评价.结果表明,该方法既保留了原影像的光谱信息,同时也改善了影像的清晰度和分辨率,利用融合后的影像进行矿区土地利用变化监测,效果明显提高.%The remote sensing image fusion method based on multi-band wavelet combining with multiple dimension texture features was discussed.In the fusion process,the high resolution panchromatic image and multi-spectral image were decomposed with multi-ary wavelet first and then local variance,local average grads,local energy and local information entropy texture features were extracted.After that the high frequency ingredients of high resolution image and R,G,B bands image by multi-band wavelet transform were fused based on combing multiple criteria of multiple dimension texture features and the high frequency ingredients of high resolution image was formed.The inverse multi-band wavelet transform was done based on low frequency ingredients of R,G,B bands image and new high frequency ingredients separately.Using 3-band wavelet transform,the fusion of 10 m resolution SPOT panchromatic and 30 m resolution TM image in Huainan mine was experimented.Visual interpretation(qualitative evaluation),spectral curve analysis and quantitative evaluation were applied to assess the method.The results show that this method can not only improve the definition and resolution of the multi-spectral image,but also retain the color feature.