共生矩阵耦合Otsu阈值的彩色图像边缘提取算法
|
摘要
针对当前彩色边缘提取算法因没有考虑不同颜色通道的关系,导致提取的边缘不连续或色彩边缘信息缺失等不足,提出了一种基于共生矩阵耦合Otsu阈值的彩色边缘提取。首先,为了评估图像的局部区域内容,计算图像中固定距离与方向上两点像素的关联,构建共生矩阵,得到了像素之间的统计特征,有效描述图像局部统计信息的变化。其次,对共生矩阵空间依次从左到右,从上到下逐像素扫描,形成属性图像。然后,利用二阶邻域来描述每个属性图像当前像素与邻域像素之间的关联,通过对当前像素与邻域的加权计算,得到8个边缘强度,并找出最大边缘强度。最后,引入Otsu阈值方法,将图像的局部最大边缘强度与最优Otsu阈值进行比较,获取其边缘像素,完成边缘提取。仿真结果表明,与当前彩色边缘提取技术相比,所提算法能够有效提取彩色边缘,获得的边缘清晰、连续性良好,取得了优异的边缘优质系数与峰值信噪比(PSNR)。
In view of the fact that the color edge extraction algorithm does not take into account the relationship between different color channels,leading to the edge discontinuity or missing information during the extraction,a color image edge extraction algorithm based on co-occurrence matrix coupled with Otsu threshold was proposed. Firstly,in order to evaluate the local region content of images,the cooccurrence matrix was constructed by analyzing the correlation between pixels in a certain distance and two directions in the algorithm image,then the statistical feature between pixels was obtained. Secondly,the co-occurrence matrix space was left to right,and the pixels were scanned from top to bottom,and then the property image was obtained by this technique. Then,using the two-order neighborhood to describe the association between each attribute of the current image pixel and its neighboring pixels,weighted by the calculation of the current pixel and its neighbor,eight edge strengthsand the maximum edge strength were obtained. Finally,the Otsu threshold method was introduced to compare the local maximum edge intensity of the image with the optimal Otsu threshold,and the edge pixels are extracted. Simulation results show that compared with the current color edge extraction technique,the proposed algorithm can efficiently extract the color edges. with clearness and continuity,and excellent edge quality coefficients and PSNR were obtained.
In view of the fact that the color edge extraction algorithm does not take into account the relationship between different color channels,leading to the edge discontinuity or missing information during the extraction,a color image edge extraction algorithm based on co-occurrence matrix coupled with Otsu threshold was proposed. Firstly,in order to evaluate the local region content of images,the cooccurrence matrix was constructed by analyzing the correlation between pixels in a certain distance and two directions in the algorithm image,then the statistical feature between pixels was obtained. Secondly,the co-occurrence matrix space was left to right,and the pixels were scanned from top to bottom,and then the property image was obtained by this technique. Then,using the two-order neighborhood to describe the association between each attribute of the current image pixel and its neighboring pixels,weighted by the calculation of the current pixel and its neighbor,eight edge strengthsand the maximum edge strength were obtained. Finally,the Otsu threshold method was introduced to compare the local maximum edge intensity of the image with the optimal Otsu threshold,and the edge pixels are extracted. Simulation results show that compared with the current color edge extraction technique,the proposed algorithm can efficiently extract the color edges. with clearness and continuity,and excellent edge quality coefficients and PSNR were obtained.
引文
[1]SWAMINATHAN A,RAMAPACKIYAM S S K.Edge detection for illumination varying images using wavelet similarity[J].IET Image Processing,2014,8(5):261-268.
[2]夏清,张振鑫,王婷婷.基于改进Sobel算子的红外图像边缘提取算法[J].激光与红外,2013,43(10):1158-1162.XIA Q,ZHANG ZH X,WANG T T.Edge extraction algorithm of on improved infrared thermal image based sobel operator[J].Laser&Infrared,2013,43(10):1158-1162.
[3]RAFSANJANI M K,VARZANEH Z A.Edge detection in digital images using ant colony optimization[J].Computer Science Journal of Moldova,2015,23(3):343-360.
[4]CHAI H,BAO T,WEE L,et al.Performance metrics for active contour models in image segmentation[J].International Journal of Physical Sciences,2014,6(27):6329-6341.
[5]龚俊,陈惠娟,郭蓉.基于四元数的彩色Canny边缘检测算法[J].南京大学学报(自然科学版),2015,51(2):279-284.GONG J,CHEN H J,GUO R.Quaternion based color Canny edge detection algorithm[J].Journal of Nanjing University(Natural Sciences),2015,51(2):279-284.
[6]杨婷,段书凯,王丽丹.基于改进忆阻细胞神经网络的彩色图像边缘提取[J].中国科学:信息科学,2017,47(7):863-877.YANG T,DUAN SH K,WANG L D.Color image edge extraction using memristor-based CNN[J].Scientia Sinica Informationis,2017,47(7):863-877.
[7]ZHUO L,HU X C,JIANG L Y.A color image edge detection algorithm based on color difference[J].Sensing and Imaging,2016,17(1):123-132.
[8]熊国萍.SUSAN边缘响应值灰度化转共生矩阵检索[J].计算机工程与应用,2016,52(5):225-230.XIONG G P.Image retrieval algorithm by GLCM through transforming SUSAN edge response matrix to gray-scale matrix[J].Computer Engineering and Applications,2016,52(5):225-230.
[9]MERY D,BERTI M A.Automatic flaw detection in aluminum castings based on the tracking of potential defects in a radioscopic image sequence[J].IEEETransaction on Robot Autom,2015,18(6):890-901.
[10]许宏科,秦严严,陈会茹.一种基于改进Canny的边缘检测算法[J].红外技术,2014,36(3):210-214.XU H K,QIN Y Y,CHEN H R.An improved algorithm for edge detection based on Canny[J].Infrared Technology,2014,36(3):210-214.
[11]张晓丹,李春来.基于离散余弦变换与共生矩阵特征的图像隐写算法[J].计算机工程,2015,41(8):127-131.ZHANG X D,LI CH L.Image steganographic algorithm based on discrete cosine transforms and co-occurrence matrix feature[J].Computer Engineering,2015,41(8):127-131.
[12]徐少平,李春泉,胡凌燕.一种改进的颜色共生矩阵纹理描述符[J].模式识别与人工智能,2013,26(1):90-98.XU SH P,LI CH Q,HU L Y.An improved color cooccurrence matrix texture descriptor[J].Pattern Recognition and Artificial Intelligence,2013,26(1):90-98.
[13]刘桂红,赵亮,孙劲光.一种改进粒子群优化算法的Otsu图像阈值分割方法[J].计算机科学,2016,43(3):309-312.Ll U G H,ZHAO L,SUN J G.Otsu image threshold segmentation method based on improved particle swarm optimization[J].Computer Science,2016,43(3):309-312.
[14]SHIH M Y,TSENG D C.Wavelet-based multiresolution edge detection and tracking[J].Image and Vision Computing,2015,23(4):441-451.
[15]SHEN R,CHENG I,BASU A.Cross-scale coefficient selection for volumetric medical image fusion[J].IEEETransactions on Bio-medical Engineering,2013,60(4):1069-1079.
[16]李伟,任青青,胡艳侠.一种复杂指针式仪表的读数自动识别算法[J].计算机技术与发展,2017,27(3):201-204.LI W,REN Q Q,HU Y X.A kind of automatic recognition algorithm for complex pointer instrument[J].Computer Technology and Development,2017,27(3):201-204.
[17]李学聪.基于机器视觉指针表检测的关键技术研究[D].广州:广东工业大学,2014:1-7.LI X C.Research on key technology of pointer table detection based on machine vision[D].Guangzhou:Guangdong University of Technology,2014:1-7.
[18]刘俊定,范新南,丁朋华.基于改进Sobel算子的多仪表图像边缘检测算法[J].科学技术与工程,2012,12(19):4691-4692.LIU J D,FAN X N,DING P H.Multi-meter image edge detection algorithm based on improved Sobel operator[J].Science Technology and Engineering,2012,12(19):4691-4692.
[2]夏清,张振鑫,王婷婷.基于改进Sobel算子的红外图像边缘提取算法[J].激光与红外,2013,43(10):1158-1162.XIA Q,ZHANG ZH X,WANG T T.Edge extraction algorithm of on improved infrared thermal image based sobel operator[J].Laser&Infrared,2013,43(10):1158-1162.
[3]RAFSANJANI M K,VARZANEH Z A.Edge detection in digital images using ant colony optimization[J].Computer Science Journal of Moldova,2015,23(3):343-360.
[4]CHAI H,BAO T,WEE L,et al.Performance metrics for active contour models in image segmentation[J].International Journal of Physical Sciences,2014,6(27):6329-6341.
[5]龚俊,陈惠娟,郭蓉.基于四元数的彩色Canny边缘检测算法[J].南京大学学报(自然科学版),2015,51(2):279-284.GONG J,CHEN H J,GUO R.Quaternion based color Canny edge detection algorithm[J].Journal of Nanjing University(Natural Sciences),2015,51(2):279-284.
[6]杨婷,段书凯,王丽丹.基于改进忆阻细胞神经网络的彩色图像边缘提取[J].中国科学:信息科学,2017,47(7):863-877.YANG T,DUAN SH K,WANG L D.Color image edge extraction using memristor-based CNN[J].Scientia Sinica Informationis,2017,47(7):863-877.
[7]ZHUO L,HU X C,JIANG L Y.A color image edge detection algorithm based on color difference[J].Sensing and Imaging,2016,17(1):123-132.
[8]熊国萍.SUSAN边缘响应值灰度化转共生矩阵检索[J].计算机工程与应用,2016,52(5):225-230.XIONG G P.Image retrieval algorithm by GLCM through transforming SUSAN edge response matrix to gray-scale matrix[J].Computer Engineering and Applications,2016,52(5):225-230.
[9]MERY D,BERTI M A.Automatic flaw detection in aluminum castings based on the tracking of potential defects in a radioscopic image sequence[J].IEEETransaction on Robot Autom,2015,18(6):890-901.
[10]许宏科,秦严严,陈会茹.一种基于改进Canny的边缘检测算法[J].红外技术,2014,36(3):210-214.XU H K,QIN Y Y,CHEN H R.An improved algorithm for edge detection based on Canny[J].Infrared Technology,2014,36(3):210-214.
[11]张晓丹,李春来.基于离散余弦变换与共生矩阵特征的图像隐写算法[J].计算机工程,2015,41(8):127-131.ZHANG X D,LI CH L.Image steganographic algorithm based on discrete cosine transforms and co-occurrence matrix feature[J].Computer Engineering,2015,41(8):127-131.
[12]徐少平,李春泉,胡凌燕.一种改进的颜色共生矩阵纹理描述符[J].模式识别与人工智能,2013,26(1):90-98.XU SH P,LI CH Q,HU L Y.An improved color cooccurrence matrix texture descriptor[J].Pattern Recognition and Artificial Intelligence,2013,26(1):90-98.
[13]刘桂红,赵亮,孙劲光.一种改进粒子群优化算法的Otsu图像阈值分割方法[J].计算机科学,2016,43(3):309-312.Ll U G H,ZHAO L,SUN J G.Otsu image threshold segmentation method based on improved particle swarm optimization[J].Computer Science,2016,43(3):309-312.
[14]SHIH M Y,TSENG D C.Wavelet-based multiresolution edge detection and tracking[J].Image and Vision Computing,2015,23(4):441-451.
[15]SHEN R,CHENG I,BASU A.Cross-scale coefficient selection for volumetric medical image fusion[J].IEEETransactions on Bio-medical Engineering,2013,60(4):1069-1079.
[16]李伟,任青青,胡艳侠.一种复杂指针式仪表的读数自动识别算法[J].计算机技术与发展,2017,27(3):201-204.LI W,REN Q Q,HU Y X.A kind of automatic recognition algorithm for complex pointer instrument[J].Computer Technology and Development,2017,27(3):201-204.
[17]李学聪.基于机器视觉指针表检测的关键技术研究[D].广州:广东工业大学,2014:1-7.LI X C.Research on key technology of pointer table detection based on machine vision[D].Guangzhou:Guangdong University of Technology,2014:1-7.
[18]刘俊定,范新南,丁朋华.基于改进Sobel算子的多仪表图像边缘检测算法[J].科学技术与工程,2012,12(19):4691-4692.LIU J D,FAN X N,DING P H.Multi-meter image edge detection algorithm based on improved Sobel operator[J].Science Technology and Engineering,2012,12(19):4691-4692.