基于等积环投影与Zernike矩的快速模板匹配
|
摘要
为增强工业级机器视觉系统的实时性和稳定性,本文提出一种基于等积环投影与Zernike矩的具有旋转不变性的快速模板匹配方法,该方法采用由粗到精匹配策略,在粗匹配阶段以等积环投影向量作为特征进行匹配得到候选点集,再基于Zenrike矩进行精确匹配.主要创新点为提出复杂度低、抗噪性强和具有旋转不变性的等积环投影向量特征,并利用圆对称性在八分圆法基础上进一步降低Zernike矩计算量.经理论分析和实验对比验证,此法在保障匹配精度前提下,其匹配速度胜于当前同类最优算法,且对高斯噪声及线性亮度变化有强鲁棒性.
In order to enhance the real-time performance and stability of industrial-level machine vision systems,this paper proposes a fast template matching method with rotation invariance based on equal-area ring projection and Zernike moments.The method uses a coarse-to-fine matching strategy.The coarse matching stage uses the equal-area ring projection vector as the feature to obtain the candidates,and then use Zenrike moments to accurately match in candidates.The main contributions of this paper are proposing the equal-are ring projection vector feature with low complexity,noise insensitive and rotation invariance,and further reducing the computational cost of Zernike moments based on the octant circle method using circular symmetry.Theoretical analysis and experimental results show that the proposed method is faster than the state-of-the-art algorithm under high matching accuracy,and it is robust to Gaussian noise and linear brightness variations.
In order to enhance the real-time performance and stability of industrial-level machine vision systems,this paper proposes a fast template matching method with rotation invariance based on equal-area ring projection and Zernike moments.The method uses a coarse-to-fine matching strategy.The coarse matching stage uses the equal-area ring projection vector as the feature to obtain the candidates,and then use Zenrike moments to accurately match in candidates.The main contributions of this paper are proposing the equal-are ring projection vector feature with low complexity,noise insensitive and rotation invariance,and further reducing the computational cost of Zernike moments based on the octant circle method using circular symmetry.Theoretical analysis and experimental results show that the proposed method is faster than the state-of-the-art algorithm under high matching accuracy,and it is robust to Gaussian noise and linear brightness variations.
引文
[1] 胡少兴,王惟达,柴进,等.基于多尺度特征的遥感图像密集匹配方法[J].光学学报,2013,33(s2):s211001-1-s211001-7.Hu Shaoxing,Wang Weida,Chai Jin,et al.Method of remote sensing images dense matching based on multi-scale features[J].Acta Optica Sinica,2013,33(s2):s211001-1-s211001-7.(in Chinese)
[2] 佟国峰,李勇,刘楠,等.大仿射场景的混合特征提取与匹配[J].光学学报,2017,37(11):1115003-1-1115003-8.Tong Guofeng,Li Yong,Liu Nan,et al.Mixed feature extraction and matching for large affine scene[J].Acta Optica Sinica,2017,37(11):1115003-1-1115003-8.(in Chinese)
[3] 王刚,孙晓亮,尚洋,等.一种基于最佳相似点对的稳健模板匹配算法[J].光学学报,2017,37(3):0315003-1-0315003-7.Wang Gang,Sun Xiaoliang,Shang Yang,et al.A robust template matching algorithm based on best-buddies similarity[J].Acta Optica Sinica,2017,37(3):0315003-1-0315003-7.(in Chinese)
[4] Wen-chia Lee,Chin-hsing Chen.A fast template matching method for rotation invariance using two stage process[A].Proceedings of the 2009 Fifth International Conference on Intelligent Information Hiding and Multimedia Signal Processing[C].Kyoto,Japan:IEEE,2009.9-12.
[5] Bohong Wei,Fan Wang,Xiaopeng Hu.Rotation-invariant template matching based on ring projection and orientation codes[A].Proceedings of the Fifth International Conference on Intelligent Control and Information Processing[C].Dalian,China:IEEE,2014.192-197.
[6] Min-seok Choi,Whoi-yul Kim.A novel two stage template matching method for rotation and illumination invariance[J].Pattern Recognition,2002,35:119-129.
[7] Yanchao Yang,Ganesh Sundaramoorthi.Shape tracking with occlusions via coarse-to-fine region based Sobolev descent[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2015,37(5):1053-1066.
[8] 尹程龙,张旭明,徐侃.基于Zernike矩的快速图像配准算法[J].光电工程,2012,39(11):81-87.YIN Cheng-long,ZHANG Xu-ming,XU Kan.A fast image registration method based on Zernike moments[J].Opto-Electronic Engineering,2012,39(11):81-87.(in Chinese)
[9] Bing C.Li.Fast Legendre moment computation for template matching[A].Proceedings of the SPIE[C].Anaheim,CA,USA:SPIE,2017.10202:102020J.
[10] 于辉,张忠秋,何周灿.用于任意旋转角度景象匹配的圆投影算法[J].计算机工程与应用,2011,47(5):172-174.YU Hui,ZHANG Zhongqiu,HE Zhoucan.Ring projection transformation algorithm for arbitrary rotation matching[J].Computer Engineering and Applications,2011,47(5):172-174.(in Chinese)
[11] Shu Huazhong,Luo Limin,Coatrieux Jean-louis.A look at moment-based approaches in imaging part 1:basic Features[J].IEEE Transactions on Engineering in Medicine and Biology Magazine(S0739-5175),2007,26(5):70-74.
[12] Shu Huazhong,Luo Limin,Coatrieux Jean-louis.A look at moment-based approaches in imaging part 2:invariance[J].IEEE Transactions on Engineering in Medicine and Biology Magazine(S0739-5175),2008,27(1):81-83.
[13] Shu Huazhong,Luo Limin,Coatrieux Jean-louis.A look at moment-based approaches in imaging part 3:computational Considerations[J].IEEE Transactions on Engineering in Medicine and Biology Magazine(S0739-5175),2008,27(3):89-91.
[14] Shu Huazhong,Luo Limin,Coatrieux Jean-louis.A look at moment-based approaches in imaging part 4:some Applications[J].IEEE Transactions on Engineering in Medicine and Biology Magazine(S0739-5175),2008,27(5):116-118.
[15] Tiansheng Wang,Simon Liao.Chinese character recognition by Zernike moments[A].Proceedings of the 2014 International Conference on Audio,Language and Image Processing[C].Shanghai,China:IEEE,2014.771-774.
[16] Amandeep Kaur,Chandan Singh.Automatic cephalometric landmark detection using Zernike moments and template matching[J].Signal Image and Video Processing,2015,9(1):117-132.
[17] Mohammed Al-Rawi.Fast Zernike moments[J].Journal of Real-Time Image Processing,2008,3:89-96.
[18] I.Area,Dimitar K.Dimitrov,E.Godoy.Recursive computation of generalised Zernike polynomials[J].Journal of Computational and Applied Mathematics,2017,312(2):58-64.
[2] 佟国峰,李勇,刘楠,等.大仿射场景的混合特征提取与匹配[J].光学学报,2017,37(11):1115003-1-1115003-8.Tong Guofeng,Li Yong,Liu Nan,et al.Mixed feature extraction and matching for large affine scene[J].Acta Optica Sinica,2017,37(11):1115003-1-1115003-8.(in Chinese)
[3] 王刚,孙晓亮,尚洋,等.一种基于最佳相似点对的稳健模板匹配算法[J].光学学报,2017,37(3):0315003-1-0315003-7.Wang Gang,Sun Xiaoliang,Shang Yang,et al.A robust template matching algorithm based on best-buddies similarity[J].Acta Optica Sinica,2017,37(3):0315003-1-0315003-7.(in Chinese)
[4] Wen-chia Lee,Chin-hsing Chen.A fast template matching method for rotation invariance using two stage process[A].Proceedings of the 2009 Fifth International Conference on Intelligent Information Hiding and Multimedia Signal Processing[C].Kyoto,Japan:IEEE,2009.9-12.
[5] Bohong Wei,Fan Wang,Xiaopeng Hu.Rotation-invariant template matching based on ring projection and orientation codes[A].Proceedings of the Fifth International Conference on Intelligent Control and Information Processing[C].Dalian,China:IEEE,2014.192-197.
[6] Min-seok Choi,Whoi-yul Kim.A novel two stage template matching method for rotation and illumination invariance[J].Pattern Recognition,2002,35:119-129.
[7] Yanchao Yang,Ganesh Sundaramoorthi.Shape tracking with occlusions via coarse-to-fine region based Sobolev descent[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2015,37(5):1053-1066.
[8] 尹程龙,张旭明,徐侃.基于Zernike矩的快速图像配准算法[J].光电工程,2012,39(11):81-87.YIN Cheng-long,ZHANG Xu-ming,XU Kan.A fast image registration method based on Zernike moments[J].Opto-Electronic Engineering,2012,39(11):81-87.(in Chinese)
[9] Bing C.Li.Fast Legendre moment computation for template matching[A].Proceedings of the SPIE[C].Anaheim,CA,USA:SPIE,2017.10202:102020J.
[10] 于辉,张忠秋,何周灿.用于任意旋转角度景象匹配的圆投影算法[J].计算机工程与应用,2011,47(5):172-174.YU Hui,ZHANG Zhongqiu,HE Zhoucan.Ring projection transformation algorithm for arbitrary rotation matching[J].Computer Engineering and Applications,2011,47(5):172-174.(in Chinese)
[11] Shu Huazhong,Luo Limin,Coatrieux Jean-louis.A look at moment-based approaches in imaging part 1:basic Features[J].IEEE Transactions on Engineering in Medicine and Biology Magazine(S0739-5175),2007,26(5):70-74.
[12] Shu Huazhong,Luo Limin,Coatrieux Jean-louis.A look at moment-based approaches in imaging part 2:invariance[J].IEEE Transactions on Engineering in Medicine and Biology Magazine(S0739-5175),2008,27(1):81-83.
[13] Shu Huazhong,Luo Limin,Coatrieux Jean-louis.A look at moment-based approaches in imaging part 3:computational Considerations[J].IEEE Transactions on Engineering in Medicine and Biology Magazine(S0739-5175),2008,27(3):89-91.
[14] Shu Huazhong,Luo Limin,Coatrieux Jean-louis.A look at moment-based approaches in imaging part 4:some Applications[J].IEEE Transactions on Engineering in Medicine and Biology Magazine(S0739-5175),2008,27(5):116-118.
[15] Tiansheng Wang,Simon Liao.Chinese character recognition by Zernike moments[A].Proceedings of the 2014 International Conference on Audio,Language and Image Processing[C].Shanghai,China:IEEE,2014.771-774.
[16] Amandeep Kaur,Chandan Singh.Automatic cephalometric landmark detection using Zernike moments and template matching[J].Signal Image and Video Processing,2015,9(1):117-132.
[17] Mohammed Al-Rawi.Fast Zernike moments[J].Journal of Real-Time Image Processing,2008,3:89-96.
[18] I.Area,Dimitar K.Dimitrov,E.Godoy.Recursive computation of generalised Zernike polynomials[J].Journal of Computational and Applied Mathematics,2017,312(2):58-64.