一种PCNN实现区域检测的图像融合方法  

Multi-Focus Image Fusion Using PCNN and Region Detection

作　　者：马洪江[1] 杨兴江[1] 向昌成[1] 

机构地区：[1]阿坝师范高等专科学校计算机科学系,四川郫县611741

出　　处：《南京师大学报（自然科学版）》2010年第3期131-134,共4页Journal of Nanjing Normal University(Natural Science Edition)

摘　　要：光学设备的有限景深造成同一场景距离不同的目标的成像效果不同,由此便产生了多聚焦图像融合问题.解决该问题的关键在于像素的分类和融合策略的选取.本文结合脉冲耦合神经网络(PCNN)模型和区域分割技术,对该问题进行尝试性研究,探讨了一种新的多聚焦图像融合算法.首先将图像的清晰度矩阵作为PCNN的输入数据,处理后得到原图像的点火映射图,然后基于映射图进行区域分割,最后根据区域分割生成融合图像.实验结果表明,该算法是有效的.Multi-focus image fusion aims at overcoming imaging cameras＇ finite depth of field by combining information from multiple images with the same scene.It＇s important for image fusion to differentiate pixel and choose the way of fusion.Pulse coupled neural network（PCNN） and region detection are employed to solve the problem of multi-focus image fusion.The clarity of original image is calculated and processed by PCNN.The original image pixels are classified based on region detection.Finally,a fusion image is reconstructed according to the classified results.Experiments indicate that the proposed method is effective.

关 键 词：图像融合 多聚焦图像 脉冲耦合神经网络 区域检测 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]