In order to improve the quality of image fusion,and preserve the spectral characteristics of the original image for avoiding the spectral degradation of the fusion image,the multi-focus image fusion algorithm based on non-down sampling Contourlet transform and region characteristic is proposed.By using the nonsubsampled Contourlet transform,multiscale decomposition is performed on the image to obtain the multi layer sub bands of the image.By using energy function of the segmentation block,the energy measurement model is built.The regional energy weighting coefficient of similarity coefficient is obtained.By using of regional energy,the similarity coefficient of the low-frequency subband is determined and the fusion of the low frequency subband image is completed.Based on the segmentation of tectonic characteristics in high frequency band,the sharpness of the ranks of summation model is formed.The high frequency subbands in the block segmentation region is obtained.Meanwhile,the sharpness value is used to establish a segmentation block decision function and complete high frequency sub image fusion.Finally,the nonsubsampled Contourlet transform is used to obtain the fusion image.The experimental results show that the image fusion algorithm is proposed is effective for multi focus image fusion.The algorithm can keep the image fusion source map in more detail,with better fusion effect.%为提高图像融合质量,较好地保留原始图像的光谱特性,避免融合图像光谱退化,提出非下采样Contourlet变换耦合区域特性多聚焦图像融合算法.采用非下采样Contourlet变换(nonsubsampled contourlet transform,NSCT)对图像进行多尺度精细分解,获取图像多层次分解子带;利用分割块区域能量函数,构造区域能量度量模型,获取区域能量相似系数,判定低频子带对应的加权系数,完成图像低频子带的融合.根据分割高频子带时形成的行列特征,形成区域锐度模型,获取高频子带分割块中的区域锐度值,利用该锐度值建立分割块判定函数,完成高频子带的融合.最后,采用非下采样Contourlet变换的逆变换得到融合图像.结果表明,与已有图像融合算法相比,本文图像融合算法融合质量更好.
展开▼
