基于自动聚焦算法的多聚焦图像融合  

Multi-focus Image Fusion Based on Automatic Focus Algorithm

作　　者：李国新[1] 王国宇[1] 王汝霖[1] 张黎[2] 

机构地区：[1]中国海洋大学电子工程系,山东青岛266071 [2]石家庄师范专科学校计算机系,河北石家庄050035

出　　处：《计算机应用研究》2005年第3期166-168,共3页Application Research of Computers

基　　金：国家自然科学基金资助项目(60272054)

摘　　要：在分析了多聚焦图像特点的基础上,将镜头自动聚焦算法移植到多聚集图像融合问题的研究中,并加以改进,其主要思想是:以高频分量作为图像清晰度的判据,寻求多聚焦图像的聚焦区域,将其直接移植到融合图像中;而对于一个焦点到另一焦点的过渡区域,对其用取最大特征的融合算子进行处理。实验表明,该算法是原图像特征的最佳近似,优于小波分解的融合算法。The paper proposes a multi-focus image fusion algorithm based on the amended automatic focus algorithm. Its main idea: first, the registered source images are divided into a lot of blocks, then using the DCT matrix of each block and the neighboring coherence in space locates the area of focus; the blocks in the focuses will be transplanted into the fused image, and the middle of area in two images between focuses will be fused by selective maximum rule. Experiments show, the fused image based on blocked DCT is the most approximation to the source images, and more than the fused result based on the WT fusion rule.

关 键 词：图像融合 多聚焦图像 自动聚焦算法 DCT变换 空间一致性 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]