基于分数阶微分差与高斯曲率滤波的边缘检测算法
|
摘要
应用梯度变化检测遥感图像纹理边缘信息时存在过检、漏检、错检和弱抗噪性等问题。为此,结合分数阶微分差和高斯曲率滤波,提出一种边缘检测算法。通过分数阶微分差运算对全色遥感图像的梯度场进行非线性增强,利用高斯曲率滤波平滑图像非线性扩散部分,并寻找正则化能量最速下降点,优化微分过程中的分数阶次和迭代次数,改善有噪图像的边缘信息提取质量。实验结果表明,该算法可抑制遥感图像纹理边缘提取过程中噪声非线性放大和扩散产生的背景伪噪声,保留图像纹理边缘信息,具有较好的图像增强和边缘检测效果。
Gradient changes are used to detect edge information of remote sensing image texture which leads to overdetection,missed detection,wrong inspection and weak noise immunity. Therefore,an edge detection algorithm is proposed by combining fractional differential difference and Gauss curvature filtering. Fractional Differential Difference Operation(FDDO) is used to realize nonlinear enhancement of gradient fields of panchromatic remote sensing images.The Gaussian curvature is used to smooth the non-linear diffusion of the image to find the fastest descent point of the regularized energy. The number of fractional orders and iterations of the differential process are optimized to improve the quality of edge information of the noisy image. Experimental results show that the algorithm can suppress background pseudo-noise caused by non-linear amplification and diffusion of noise in the process of texture edge extraction of remote sensing images,and retain texture edge information. It has better image enhancement and edge detection effect.
Gradient changes are used to detect edge information of remote sensing image texture which leads to overdetection,missed detection,wrong inspection and weak noise immunity. Therefore,an edge detection algorithm is proposed by combining fractional differential difference and Gauss curvature filtering. Fractional Differential Difference Operation(FDDO) is used to realize nonlinear enhancement of gradient fields of panchromatic remote sensing images.The Gaussian curvature is used to smooth the non-linear diffusion of the image to find the fastest descent point of the regularized energy. The number of fractional orders and iterations of the differential process are optimized to improve the quality of edge information of the noisy image. Experimental results show that the algorithm can suppress background pseudo-noise caused by non-linear amplification and diffusion of noise in the process of texture edge extraction of remote sensing images,and retain texture edge information. It has better image enhancement and edge detection effect.
引文
[1]魏巍,吴孔平,郭来功,等.基于联合非负字典学习的遥感图像超分辨重建[J].计算机工程,2016,42(8):271-276.
[2]杨柱中,周激流,晏祥玉,等.基于分数阶微分的图像增强[J].计算机辅助设计与图形学学报,2008,20(3):343-348.
[3]薄亦非.将分数微分演算引入数字图像处理[J].四川大学学报(工程科学版),2007,39(3):124-132.
[4]GONG Yuanhao,SBALZARINI I F.Curvature filters efficiently reduce certain variational energies[J].IEEE Transactions on Image Processing,2017,26(4):1786-1798.
[5]MATHIEU B,MELCHIOR P,OUSTALOUP A,et al.Fractional differentiation for edge detection[J].Signal Processing,2003,83(11):2421-2432.
[6]陈青,刘金平,唐朝晖,等.基于分数阶微分的图像边缘细节检测与提取[J].电子学报,2013,41(10):1873-1880.
[7]程金梅,叶永强,姜斌.利用复合导数的边缘检测新算法[J].中国图象图形学报,2012,17(3):393-401.
[8]蒲亦非,王卫星.数字图像的分数阶微分掩模及其数值运算规则[J].自动化学报,2007,33(11):1128-1135.
[9]韩毅,赵凯,周晏.基于分数阶变换和改进最小生成树的图像配准算法[J].计算机工程,2017,43(9):263-269.
[10]杨柱中,周激流,黄梅,等.基于分数阶微分的边缘检测[J].四川大学学报(工程科学版),2008,40(1):152-157.
[11]成宽洪,周慧鑫,秦翰林,等.基于曲率滤波和梯度变换的图像增强[J].光子学报,2017,46(7):159-164.
[12]ABDOU I E,PRATT W.Quantitative design and evaluation of enhancement/thresholding edge detectors[J].Proceedings of the IEEE,1979,67(5):753-763.
[13]GONG Y,LAZEBNIK S,GORDO A,et al.Iterative quantization:a procrustean approach to learning binary codes for large-scale image retrieval[J].IEEETransactions on Pattern Analysis and Machine Intelligence,2013,35(12):2916-2929.
[14]DAVIS J,GOADRICH M.The relationship between precision-recall and ROC curves[C]//Proceedings of International Conference on Machine Learning.New York,USA:ACM Press,2006:233-240.
[15]FAWCETT T.An introduction to ROC analysis[J].Pattern Recognition Letters,2006,27(8):861-874.
[2]杨柱中,周激流,晏祥玉,等.基于分数阶微分的图像增强[J].计算机辅助设计与图形学学报,2008,20(3):343-348.
[3]薄亦非.将分数微分演算引入数字图像处理[J].四川大学学报(工程科学版),2007,39(3):124-132.
[4]GONG Yuanhao,SBALZARINI I F.Curvature filters efficiently reduce certain variational energies[J].IEEE Transactions on Image Processing,2017,26(4):1786-1798.
[5]MATHIEU B,MELCHIOR P,OUSTALOUP A,et al.Fractional differentiation for edge detection[J].Signal Processing,2003,83(11):2421-2432.
[6]陈青,刘金平,唐朝晖,等.基于分数阶微分的图像边缘细节检测与提取[J].电子学报,2013,41(10):1873-1880.
[7]程金梅,叶永强,姜斌.利用复合导数的边缘检测新算法[J].中国图象图形学报,2012,17(3):393-401.
[8]蒲亦非,王卫星.数字图像的分数阶微分掩模及其数值运算规则[J].自动化学报,2007,33(11):1128-1135.
[9]韩毅,赵凯,周晏.基于分数阶变换和改进最小生成树的图像配准算法[J].计算机工程,2017,43(9):263-269.
[10]杨柱中,周激流,黄梅,等.基于分数阶微分的边缘检测[J].四川大学学报(工程科学版),2008,40(1):152-157.
[11]成宽洪,周慧鑫,秦翰林,等.基于曲率滤波和梯度变换的图像增强[J].光子学报,2017,46(7):159-164.
[12]ABDOU I E,PRATT W.Quantitative design and evaluation of enhancement/thresholding edge detectors[J].Proceedings of the IEEE,1979,67(5):753-763.
[13]GONG Y,LAZEBNIK S,GORDO A,et al.Iterative quantization:a procrustean approach to learning binary codes for large-scale image retrieval[J].IEEETransactions on Pattern Analysis and Machine Intelligence,2013,35(12):2916-2929.
[14]DAVIS J,GOADRICH M.The relationship between precision-recall and ROC curves[C]//Proceedings of International Conference on Machine Learning.New York,USA:ACM Press,2006:233-240.
[15]FAWCETT T.An introduction to ROC analysis[J].Pattern Recognition Letters,2006,27(8):861-874.