摘要
为了解决当前遥感图像融合算法因忽略了区域中像素点的边缘特征而导致融合图像中存在块效应以及模糊效应的不足,在非下采样Shearlet变换的基础上,设计了基于边缘制约模型的遥感图像融合算法。首先,将多光谱(MS)图像经过IHS分解,提取相应的亮度分量。然后,通过非下采样Shearlet变换,将全色(PAN)图像与亮度分量进行分解,获取各自的高频系数与低频系数。再通过图像的空间频率特征,建立低频系数的融合函数,对低频系数进行融合。并利用图像的区域平均梯度特征与图像区域中像素点的边缘能量特征,构造了边缘制约模型,对高频系数进行融合。最后,将融合后低频系数、高频系数经非下采样Shearlet逆变换和IHS逆变换,获取融合图像。实验结果显示,与当前遥感图像融合方法相比,所提算法的融合图像具有更高的清晰度,更好地保持了图像的光谱特性,消除了块效应以及模糊效应。
In order to solve the problem of blocking effect and blurring effect induced by ignoring the edge features of image region pixels t in current remote sensing image fusion algorithms. the remote sensing image fusion algorithm based on edge constraint model and nonsubsampled shearlet transform was proposed in this paper. Firstly,multispectral images are decomposed by luminance hue saturation decomposition to extract luminance components. Then, the panchromatic image and luminance components are decomposed by nonsubsampled Shearlet transform to obtain high-frequency coefficients and low-frequency coefficients. Finally,the fusion function of lowfrequency coefficients is established through the spatial frequency characteristics of the image to fuse the low-frequency coefficients. An edge constraint model is constructed to fuse the high frequency coefficients by using the average gradient feature and the edge energy feature of pixels in the image region. After the fusion,the low-frequency coefficients and high-frequency coefficients are inversely transformed by non-downsampling Shearlet,then the fused images are obtained by inverse transform of IHS. The experimental results show that,compared with the current remote sensing image fusion methods,the fusion images designed in this paper not only have better clarity,but also have better spectral characteristics without blocking effect and blurring effect.
引文
[1] WANG J,PENG J Y,JIANG X Y. Remote-sensingimage fusion using sparse representation with sub-dictionaries[J]. International Journal of RemoteSensing,2017,38(12):3564-3585.
[2] ZHANG L B,ZHANG J. A new saliency-driven fusionmethod based on complex wavelet transform for remotesensing images[J].IEEE Geoscience and Remote SensingLetters,2017,14(12):2433-2437.
[3] TIAN B H,LAN L,SHI H L. Remote sensing image fusionscheme using directional vector in NSCT domain[J].Telkomnika Telecommunication,Computing,Electronicsand Control,2016,14(2):598-606.
[4]徐金东,倪梦莹,童向荣.一种基于多尺度稀疏分解的遥感图像融合新方法[J].国土资源遥感,2017,29(3):51-58.XU J D,NI M Y,TONG X R. A new method for remotesensing image fusion based on multi-scale sparsedecomposition[J]. Remote Sensing for Land&Resources,2017,29(3):51-58.
[5] MIAO Q G,LIU R Y,QUAN Y N. Remote sensingimage fusion based on shearlet and genetic algorithm[J].International Journal of Bio-Inspired Computation,2017,9(4):240-250.
[6] DENG C,WANG Z H,LI X W. An improved remotesensing image fusion algorithm based on ihstransformation[J]. Ksii Transaction on Internet andinformation Systems,2017,11(3):1633-1649.
[7] ZHU X L,BAO W X. Comparison of remote sensingimage fusion strategies adopted in HSV and IHS[J].Journal of the Indian Society of Remote Sensing,2018,46(3):377-385.
[8]赵学军,刘静.基于Shearlet和稀疏表示的遥感图像融合[J].科学技术与工程,2017,17(4):255-259.ZHAO X J,LIU J. Remote sensing image fusion based onshearlet and sparse representation[J].Science Technologyand Engineering,2017,17(4):255-259.
[9] LUO X Q,ZHANG Z C,WU X J. A novel algorithm ofremote sensing image fusion based on shift-invariantShearlet transform and regional selection[J]. AEU:International Journal of Electronics&Communications,2016,70(2):186-197.
[10] WAN W G, YANG Y, LEE H J. Practical remotesensing image fusion method based on guided filter andimproved SML in the NSST domain[J]. Signal,Imageand Video Processing,2018,12(5):959-966.
[11]杨成立,殷鸣,蒋红海.基于非下采样Shearlet变换的磁瓦表面裂纹检测[J].农业机械学报,2017,48(3):405-412.YANG CH L,YIN M,JIANG H H. Detection of surfacecrack defects in magnetic tile images based onnonsubsampled Shearlet transform[J].Transactions of theChinese Society for Agricultural Machinery, 2017,48(3):405-412.
[12]董崧,臧淑英,吴长山.SPOT-7遥感图像融合技术对比研究[J].测绘与空间地理信息,2017,40(1):75-78.DONG S,ZANG SH Y,WU CH SH. Comparison ofdifferent image fusion methods using SPOT-7[J].Geomatics&Spatial Information Technology, 2017,40(1):75-78.
[13] TANG S Z, SHEN C M, ZHANG G X. Adaptiveregularized scheme for remote sensing image fusion[J].Frontiers of Earth Science,2016,10(2):236-244.
[14] WEI J B,HUANG Y K. NMPE:A normalized metric formeasuring generalized spatial distortion of multispectralpanshapening fusion[J]. Multimedia Tools andApplications,2017,76(21):23099-23116.
[15] LI F. Digital signal transformation and computer imageprocessing and analysis based on multi-sensor imagefusion[J]. 6th International Conference on Machinery,Materials,Environment,Biotechnology and Computer,2016,88(1):399-404.
[16] LIU S Q,SHI M Z,ZHU Z H. Image fusion based oncomplex-shearlet domain with guided filtering[J].Multidimensional Systems and Signal Processing,2017,28(1):207-224.
[17]纪峰,李泽仁,常霞.基于PCA和NSCT变换的遥感图像融合方法[J].图学学报,2017,38(2):247-252.JI F,LI Z R,CHANG X. Remote sensing image fusionmethod based on PCA and NSCT transform[J]. Journalof Graphics,2017,38(2):247-252.
[18] BAO W X,WANG W,ZHU Y X. Pleiades satelliteremote sensing image fusion algorithm based on shearlettransform[J]. Journal of the Indian Society of RemoteSensing,2018,46(1):19-29.
[19] LIU G,LI L,GONG H. Multisource remote sensingimagery fusion scheme based on bidimensional empiricalmode decomposition and Its application to the extractionof bamboo forest[J]. Remote Sensing,2017,9(1):19-26.
[20]李红,刘芳,张凯.稀疏非负矩阵分解下的遥感图像融合[J].西安电子科技大学学报,2016,43(2):193-198.LI H,LIU F,ZHANG K. Remote sensing image fusionbased on sparse non-negative matrix factorization[J].Journal of Xidian University,2016,43(2):193-198.