摘要
针对当前图像融合算法因利用下采样分解导致信息冗余、融合图像对比度和清晰度不高等不足,提出基于非分离小波多方向分析NSWMDA(non-separable wavelet multi-directional analysis)与脉冲耦合神经网络PCNN(pulse coupled neural network)的图像融合算法。根据非分离小波变换和非下采样方向滤波器,设计多尺度NSWMDA机制,获取源图像的低频、高频子带;引入高斯-拉普拉斯算子,提取高频子带的高频信息,设计自适应PCNN融合准则,完成高频信息融合;基于高斯加权平均的低频融合规则,对低频子带完成融合;利用NSWMDA重构,输出融合图像。仿真结果表明,与当前图像融合算法相比,该算法的融合图像具有更优的交互信息、PSNR(peak signal to noise ratio),以及融合效率。
Because insufficient sampling decomposition leads to redundant information and the contrast and clarity of fusion image are poor in the current image fusion algorithms,the image fusion algorithm based on NSWMDA coupling adaptive PCNN was proposed.Based on non separable wavelet transform and non down sampling direction filter,a multi scale NSWMDA mechanism was designed to obtain the low frequency sub bands and high frequency ones of the source image.The Laplasse Gauss operator was introduced to extract the high frequency sub band information,and the high frequency fusion criterion of adaptive PCNN was designed.The low frequency fusion based on Gauss weighted average was adopted.The high frequency and low frequency information were processed using NSWMDA transform,the fusion image was synthesized.Results of experiments show contrasted with the current image fusion algorithm,using the proposed method can maintain better mutual information of the fused image,and improve the peak signal to noise ratio as well as the fusion efficiency.
作者
汪小威
陆涛
刘伟晖
龚向欣
WANG Xiao-wei LU Tao LIU Wei-hui GONG Xiang-xin(College of Information Engineering, Nanning University, Nanning 530200,China College of Computer and Electronic Information, Guangxi University, Nanning 530004,China College of Computer Science and Engineering, Nanjing University of Science and Technology, Nanjing 210094,China)
出处
《计算机工程与设计》
北大核心
2017年第11期3118-3124,共7页
Computer Engineering and Design
基金
2015年广西省高校重点科研基金项目(KY2015Y13530)
广西自然科学基金项目(2014GXJSFAA019272)
