基于改进局部梯度的小波图像融合方法  

New Image Fusion Method based-on Improved Local Gradient Processing of Wavelet Coefficient

作　　者：曾基兵[1] 陈怀新[2] 王卫星[1] 

机构地区：[1]电子科技大学,四川成都610054 [2]中国西南电子技术研究所,四川成都610036

出　　处：《电视技术》2007年第B08期18-20,共3页Video Engineering

摘　　要：给出了一种基于双正交小波交换的图像融合方法。针对图像小波分解后的高频系数,提出了基于局部梯度的选择和加权平均相结合的系数融合规则,在充分保留源图像细节信息前提下,保证融合结果的一致性。对多聚焦图像进行试验,并用交叉熵等参数作客观评价,结果表明本方法所得的融合图像能反映更丰富的细节信息,具有更强的噪声抑制能力和更好的视觉效果。A multi-sensor image fusion algorithm using biorthogonal wavelet transform is presented. In allusion to the high-frequency coefficients of the wavelet after decomposition, a fusion rule based-on the selection of local gradient and weighted average is proposed. This preserves both the details of source image information and ensures the consistency of fusion results. The algo- rithm is used to fuse multi-focus images and the quantitative assessment is also presented using the evaluation parameter of cross entropy and so on. The results show that this algorithm can preserve richer details from the primitive image information, and has stronger ability of restraining noise and better visual effect.

关 键 词：图像融合 小波分解 改进局部梯度 融合性能 

分 类 号：TN911.73[电子电信—通信与信息系统]