Abstract: A novel algorithm of image fusion based on the regional features of wavelet coefficients is proposed.After the original images being decomposed by wavelet transform，low-frequency and high-frequency coefficients in different scales and various direction are obtained.Then，the low-frequency coefficients of fusion images are determined by using the correlation coefficient and the regional variance，while the high-frequency coefficients are determined by using local regional energy.Finally，the fused image is reconstructed by inverse wavelet transform.Many groups of images are taken as experimental data，and fusion results are evaluated by average gradient，information entropy and spatial frequency.Experimental results show that the proposed algorithm is more outstanding than the conventional methods and has better fusion effects.

Key words: image fusion, wavelet transform, regional feature, correlation coefficient, regional energy

摘要： 提出了一种基于区域特性的小波图像融合新算法。对原始图像进行小波多尺度分解，得到在不同尺度和方向下的低频系数和高频系数;对低频系数，采用图像区域之间的相关系数和区域方差的融合规则确定低频融合系数，而对不同尺度和方向下的高频系数，采用基于局部区域能量的融合规则确定高频融合系数;最后，通过小波逆变换得到融合图像。对多组图像进行了融合仿真实验，并用平均梯度、信息熵和空间频率对融合结果进行了客观评价。实验结果表明，该算法优于传统的融合算法，取得了更好的融合效果。

关键词: 图像融合, 小波变换, 区域特性, 相关系数, 区域能量