一种基于深度神经网络模型的多聚焦图像融合方法

被引:6
作者
刘帆
陈泽华
柴晶
机构
[1] 太原理工大学信息工程学院
基金
山西省青年科学基金;
关键词
多聚焦图像; 图像融合; 小波核滤波器; 深度神经网络; 自动编码器;
D O I
暂无
中图分类号
TP391.41 []; TP183 [人工神经网络与计算];
学科分类号
080203 ;
摘要
基于多聚焦图像融合中存在的低频信息易产生缺失的现象进行分析,提出一种基于深度神经网络模型的低频子带融合策略,并结合小波核滤波器及针对高频子带的融合策略,给出多聚焦图像融合方法。该方法利用自动编码器提取低频子带特征,利用网络隐层中的权值信息选择低频子带分量。采用3组聚焦不同的自然图像及1组医学图像进行算法测试,并与传统的低频子带融合策略进行对比,同时比较基于轮廓波变换的多聚焦图像融合方法、基于非下采样轮廓波变换的多聚焦图像融合方法。试验结果表明:其中一组图像采用深度神经网络模型的策略所得到的融合结果的边缘融合指标值能够达到0.802 7,优于其余比较方法的0.761 4、0.722 7和0.716 4,从而证实基于深度神经网络模型的融合策略的有效性。
引用
收藏
页码:7 / 13
页数:7
相关论文
共 25 条
  • [1] Compressed-domain ship detection on spaceborne optical image using deep neural network and extreme learning machine. TANG J X,DENG C W,HUANG G B,et al. IEEE Transactions on Geoscience and Remote Sensing . 2015
  • [2] A new quality metric for image fusion. Piella G,,Heijmans H. IEEE International Conference on ImageProcessing . 2003
  • [3] A Wavelet-based Scene Image Fusion Algorithm. HUANG Xish,CHEN Zhe. IEEE . 2002
  • [4] Deep embedding netw ork for clustering. HUANG P H,HUANG Y,WANG W,et al. 22nd International Conference on Pattern Recognition . 2014
  • [5] SAR image despeckling based on w avelet kernel transform and gaussian scale mixture model. LIU F,JIAO L C,YANG S Y. 2nd Asian-Pacific Conference on Synthetic Aperture Radar . 2009
  • [6] EEG-based emotion classification using deep belief networks. Zheng W.L,Zhu J.Y,Peng Y.et al. 2014 IEEE International Conference on Multimedia and Expo . 2014
  • [7] Learning deep architectures for AI. Bengio,Yoshua. Foundations and Trends in Machine Learning . 2009
  • [8] Greedy layer-wise training of deepnetworks. Bengio Yoshua,,Lamblin Pascal,Popovici Dan, et al. Advances in Neural Information Processing Systems . 2007
  • [9] Multi-Focus Image Fusion based on Stationary WaveletTransfom and Extended Spatial Frequency Measurement. Borwonwatanadelok P,Rattanapitak W,Udomhunsakul S. International Conference on Electronic Computer Technolo-gy . 2009
  • [10] Review article Multisensor image fusion in remote sensing: concepts, methods and applications[J] . C. Pohl,J. L. Van Genderen. &nbspInternational Journal of Remote Sensing . 1998 (5)