基于三通道不可分对称小波的多聚焦图像融合  

Multi-focus image fusion method based on three channels non-separable symmetrical wavelets

作　　者：黄倩文 刘斌[1] 李庄[2] 

机构地区：[1]湖北大学数学与计算机科学学院,湖北武汉430062 [2]武汉科技大学信息科学与工程学院,湖北武汉430081

出　　处：《湖北大学学报（自然科学版）》2012年第4期444-448,457,共6页Journal of Hubei University：Natural Science

基　　金：国家自然科学基金(61072126)资助

摘　　要：提出一种基于三通道不可分对称小波的多聚焦图像融合新算法.利用矩阵扩充的方法,给出一种三通道不可分对称小波滤波器组的构造方法,用所构造的滤波器分别对两组多聚焦图像作非下采样多尺度分解,对低频子带采用区域能量取大、对高频子带采取区域锐度取大的融合规则来对分解后的子图像进行融合,最后通过小波逆变换得到融合结果图像.实验结果表明,基于该方法的融合结果图像比传统的基于张量积小波的方法具有更高的空间分辨率和清晰度.Put forward a new algorithm that based on the three-channel inseparable symmetrical wavelets about multi-focus image fusion.Using the method of matrix expansion,a construction method of the symmetrical filter banks relating to this kind of wavelet was proposed.The multi-resolution decompositions of the multi-focus were performed in non-subsampled mode by using the filter banks constructed.The fusion rule of selecting the bigger weighted energy of the low frequency sub-band coefficients and the bigger average gradient of the high frequency sub-band coefficients were employed.Finally the fused image was achieved through inverse non-separable wavelet transform.The experimental results showed that the method had good visual effect and the fused images had rich information,high spatial resolution and clarity.The fusion performance better than traditional method based on DWT.

关 键 词：图像融合 不可分小波 区域能量 区域锐度 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]