人工鱼群算法优化的小波图像融合  

Artificial Fish Algorithm Using in Wavelet Image Fusion

作　　者：郑伟[1] 张京谱 田华[1] 孙雪青[1] 

机构地区：[1]河北大学电子信息工程学院,河北保定071002

出　　处：《电视技术》2014年第1期26-29,33,共5页Video Engineering

基　　金：河北大学医工交叉研究中心开放基金项目(BM201103)

摘　　要：针对甲状腺肿瘤的B超、核素图像和颅腔横断面的CT,MRI图像,提出一种使用人工鱼群算法寻优的小波图像融合方法。利用小波变换将源图像进行小波分解,得到源图像的低频部分和高频部分。低频部分的系数使用加权融合算法进行融合,为使融合结果包含足够的信息量,以综合熵作为适应值对低频部分使用人工鱼群算法进行优化,得到融合系数的最优权值。高频部分的系数采用取绝对值最大的融合规则。通过小波逆变换得到融合图像。实验过程中,分别对这两组医学图像进行融合。客观评价参数表明该算法能够得到更优的融合图像。According to B ultrasonic,nuclide images of thyroid tumor and CT,MRI images of cranial cavity,this paper puts forward a fusion method which using artificial fish-swam algorithm based on wavelet image fusion. Apply wavelet transform to source images giving high frequency sub-bands and low frequency sub-band. Fusing the low frequency sub-band with the method of weighted fusion algorithm. In order to make the fusion results contain e- nough information ,choose the associated entropy as the adaptive value of artificial fish-swam algorithm and use the algorithm to obtain the optimal weights of the low frequency sub-band. And the high frequency sub-bands are fused by the rule of maximum absolute value. The fusion image can be obtained by the wavelet inverse transformation. Fusing the two groups of medical images with the fusion algorithm,the objective evaluation parameter shows that the fusion algorithm can get excellent fusion images.

关 键 词：图像融合 小波变换 人工鱼群算法 医学图像 

分 类 号：TN911.73[电子电信—通信与信息系统] TP391.4[电子电信—信息与通信工程]