大豆冠层多源图像特征点配准方法研究
|
摘要
针对具有颜色信息的大豆冠层三维结构形态的重建问题,采用PMD摄像机与彩色摄像机相结合的多源图像采集系统获取大豆冠层多源图像,对大豆冠层多源图像特征点配准方法进行研究。以彩色图像和强度图像为研究对象,利用仿射变换实现彩色图像坐标系到PMD图像坐标系的转换;利用Harris算法检测图像特征点,采用基于归一化互相关系数法(NCC)实现特征点粗匹配。为克服传统RANSAC算法抽样次数较多及和数据检验时间较长的弊端,提出在特征点匹配阶段,按照可信度将特征点对排序,从可信度高的点对开始抽取的方法来优化经典RANSAC算法,进而实现特征点精匹配,最终完成多源图像特征点配准。为验证本研究提出的图像配准算法的有效性,将该算法与传统图像配准算法相对比,结果表明:室外和室内环境下,样本组的平准正确配准率分别为83%和87%,均优于传统图像配准算法,并满足快速配准大豆冠层多源图像特征点的要求。
In order to reconstruct the three-dimensional structure of soybean canopy with color information,a multisource image acquisition system combining PMD camera and color camera was used to acquire the multi-source image of soybean canopy,and the feature points of multi-source images of soybean canopy were studied.By using color image and intensity images as research objects,affine transformation was exploited to realize the transformation from color image coordinate system to PMD image coordinate system firstly,Harris algorithm was then to test the image feature points.Normalized cross correlation was adopted to achieve coarse matching for feature points.In addition,for the purpose of overcoming disadvantages of classic RANSAC algorithm,such as more sampling times and longer time of data validation,an improved RANSAC algorithm was used to realize precise matching,which was based on classification by credibility during stage of matching feature points.We finally accomplished Feature point registration of multi-source image after finishing the steps mentioned above was accomplished.To verify the validity of the image registration algorithm proposed in this study,algorithms of this article were compared to with the traditional image registration algorithm.The test results showed that the correct registration ratios of the sample groups were 83% and87%,respectively under the circumstance of outside and inside environments,and both of them were better than the classical image registration algorithms,which could meet the need of quick registering feature points of multi-source images of soybean canopies.
In order to reconstruct the three-dimensional structure of soybean canopy with color information,a multisource image acquisition system combining PMD camera and color camera was used to acquire the multi-source image of soybean canopy,and the feature points of multi-source images of soybean canopy were studied.By using color image and intensity images as research objects,affine transformation was exploited to realize the transformation from color image coordinate system to PMD image coordinate system firstly,Harris algorithm was then to test the image feature points.Normalized cross correlation was adopted to achieve coarse matching for feature points.In addition,for the purpose of overcoming disadvantages of classic RANSAC algorithm,such as more sampling times and longer time of data validation,an improved RANSAC algorithm was used to realize precise matching,which was based on classification by credibility during stage of matching feature points.We finally accomplished Feature point registration of multi-source image after finishing the steps mentioned above was accomplished.To verify the validity of the image registration algorithm proposed in this study,algorithms of this article were compared to with the traditional image registration algorithm.The test results showed that the correct registration ratios of the sample groups were 83% and87%,respectively under the circumstance of outside and inside environments,and both of them were better than the classical image registration algorithms,which could meet the need of quick registering feature points of multi-source images of soybean canopies.
引文
[1]朱冰琳,刘扶桑,朱晋宇,郭焱,马韫韬.基于机器视觉的大田植株生长动态三维定量化研究[J].农业机械学报,2018,49(5):256-262Zhu B L,Liu F S,Zhu J Y,Guo Y,Ma Y T.Three-dimensional quantifications of plant growth dynamics in field-grown plants based on machine vision method[J].Transactions of the Chinese Society for Agricultural Machinery,2018,49(5):256-262(in Chinese)
[2] Xu J,Qi D W.The research of tree growth based on image vision theory[C].In:International Conference on Image Analysis and Signal Processing 2011.Prague:IEEE,2011:244-247
[3] Nielsen M,Slaughter D C,Gliever C.Vision-based 3Dpeach tree reconstruction for automated blossom thinning[J].IEEE Transactions on Industrial Informatics,2012,8(1):188-196
[4]王菲.高纺锤形富士苹果树分形维数及三维数字化建模和STAR值的研究[D].杨凌:西北农林科技大学,2012Wang F.Research on the fractal dimension and threedimensional digitized modeling and star value of‘FUJI’apple trees trained to tall spindle shaped[D].Yangling:Northwest A&F University,2012(in Chinese)
[5] Watanabe T,Hanan J S,Room P M.Rice morphogenesis and plant architecture:Measurement. specification and the reconstruction of structural develop ment by 3Darchitectural modelling[J].Annals of Botany,2005,95(7):1131-1143
[6] Zhu J J,Wang L,Yang R G,Davis J E,Pan Z G.Reliability fusion of time-of-flight depth and stereo geometry for high quality depth maps[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2011,33(7):1400-1414
[7] Golbach F,Kootstra G,Damjanovic S,Otten G,van de Zedde R. Validation of plant part measurements using a 3D reconstruction method suitable for high-throughput seedling phenotyping[J].Machine Vision and Applications,2016,27(5):663-680
[8]张旭东,沈玉亮,胡良梅,陈菁菁.改进的PMD距离图像超分辨率重建算法[J].中国图像图形学报,2012,17(4):480-486Zhang X D,Shen Y L,Hu L M,Chen J J.Improved superresolution reconstruction algorithm for PMD range image[J].Journal of Image and Graphics,2012,17(4):480-486(in Chinese)
[9]黄宝康.基于灰度的图像配准技术研究[D].赣州:江西理工大学,2016Huang B K,Research on intensity-based image registration technology[D].Ganzhou:Jiangxi Univesity of Science and Technology(in Chinese)
[10]冯骥.基于互信息和B样条FFD模型的图像配准算法研究[D].哈尔滨:哈尔滨理工大学,2016Feng J.Research of image registration based on mutual information and b-spline FFD model[J].Harbin:Harbin University of Science and Technology,2016(in Chinese)
[11]冯娟,刘刚,王圣伟,马晓丹,周薇.采摘机器人果实识别的多源图像配准[J].农业机械学报,2013,44(3):197-203Feng J,Liu G,Wang S W,Ma X D,Zhou W.Multi-source images registration for harvesting robot to recognize fruits[J].Transactions of the Chinese Society for Agricultural Machinery,2013,44(3):197-203(in Chinese)
[12]马晓丹,刘刚,冯娟,周薇.成熟期苹果树冠层器官异源图像配准[J].农业机械学报,2014,45(4):82-88,140Ma X D,Liu G,Feng J,Zhou W. Multi-source image registration for canopy organ of apple trees in mature period[J].Transactions of the Chinese Society for Agricultural Machinery,2014,45(4):82-88,140(in Chinese)
[13] Wang F B,Tu P,Wu C,Chen L,Feng D.Multi-image mosaic with SIFT and vision measurement for microscale structures processed by femtosecond laser[J].Optics and Lasers in Engineering,2018,100:124-130
[14] Pan L,Shi F,Zhu W,Guan L,Chen X.Detection and registration of vessels for longitudinal 3Dretinal OCT images using SURF[C].In:Medical Imaging 2018:Biomedical Applications in Molecular,Structural,and Functional Imaging.Houston:SPIE,2018,10578:105782
[15]毛雁明,兰美辉,王运琼,冯乔生.一种改进的基于Harris的角点检测方法[J].计算机技术与发展,2009,19(5):130-133Mao Y M,Lan M H,Wang Y Q,Feng Q S.An improved corner detection method based on harris[J].Computer Technology and Development,2009,19(5):130-133(in Chinese)
[16] Byravan A,Fox D.Se3-nets:Learning rigid body motion using deep neural networks[C].In:IEEE International Conference on Robotics and Automation(ICRA)2017.Singapore:IEEE,2017:173-180
[17] Efimov A I,Novikov A I.An algorithm for multistage projective transformation adjustment for image superimposition[J].Computer Optics,2016,40(2):258-265
[18] Kuroki Y,Takenaka K,Motomatsu S.Robust keypoint detection against affine transformation using moment invariants on intrinsic mode function[C].In:International Conference on Intelligent Informatics and Biomedical Sciences(ICIIBMS)2015.Okinawa:IEEE,2015:403-407
[19] Zhong Z,Li J,Luo ZM,Chapman M.Spectral-spatial residual network for hyperspectral image classification:a 3-D deep learning framework[J].IEEE Transactions on Geoscience and Remote Sensing,2018,56(2):847-858
[20] Du H,Xu Z,Ding Y.The fast lane detection of road using RANSAC algorithm[C].In:International Conference on Applications and Techniques in Cyber Security and Intelligence.Ningbo:Edizioni della Normale,2017:1-7
[21]黄梅.基于改进RANSAC算法的图像拼接技术[J].海南大学学报:自然科学版,2011,29(2):172-177Huang M.Image mosaic method based on improvedRANSAC[J].Journal of HaiNan University:Natural Science,2011,29(2):172-177(in Chinese)
[22]吴恩生,朱敏琛.一种融合局部与全局信息的距离约束角点匹配算法[J].计算机应用,2010,30(1):68-70,81Wu E S,Zhu M C.Corner matching method of constraints of distance combining local and global information[J].Journal of Computer Applications,2010,30(1):68-70,81(in Chinese)
[2] Xu J,Qi D W.The research of tree growth based on image vision theory[C].In:International Conference on Image Analysis and Signal Processing 2011.Prague:IEEE,2011:244-247
[3] Nielsen M,Slaughter D C,Gliever C.Vision-based 3Dpeach tree reconstruction for automated blossom thinning[J].IEEE Transactions on Industrial Informatics,2012,8(1):188-196
[4]王菲.高纺锤形富士苹果树分形维数及三维数字化建模和STAR值的研究[D].杨凌:西北农林科技大学,2012Wang F.Research on the fractal dimension and threedimensional digitized modeling and star value of‘FUJI’apple trees trained to tall spindle shaped[D].Yangling:Northwest A&F University,2012(in Chinese)
[5] Watanabe T,Hanan J S,Room P M.Rice morphogenesis and plant architecture:Measurement. specification and the reconstruction of structural develop ment by 3Darchitectural modelling[J].Annals of Botany,2005,95(7):1131-1143
[6] Zhu J J,Wang L,Yang R G,Davis J E,Pan Z G.Reliability fusion of time-of-flight depth and stereo geometry for high quality depth maps[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2011,33(7):1400-1414
[7] Golbach F,Kootstra G,Damjanovic S,Otten G,van de Zedde R. Validation of plant part measurements using a 3D reconstruction method suitable for high-throughput seedling phenotyping[J].Machine Vision and Applications,2016,27(5):663-680
[8]张旭东,沈玉亮,胡良梅,陈菁菁.改进的PMD距离图像超分辨率重建算法[J].中国图像图形学报,2012,17(4):480-486Zhang X D,Shen Y L,Hu L M,Chen J J.Improved superresolution reconstruction algorithm for PMD range image[J].Journal of Image and Graphics,2012,17(4):480-486(in Chinese)
[9]黄宝康.基于灰度的图像配准技术研究[D].赣州:江西理工大学,2016Huang B K,Research on intensity-based image registration technology[D].Ganzhou:Jiangxi Univesity of Science and Technology(in Chinese)
[10]冯骥.基于互信息和B样条FFD模型的图像配准算法研究[D].哈尔滨:哈尔滨理工大学,2016Feng J.Research of image registration based on mutual information and b-spline FFD model[J].Harbin:Harbin University of Science and Technology,2016(in Chinese)
[11]冯娟,刘刚,王圣伟,马晓丹,周薇.采摘机器人果实识别的多源图像配准[J].农业机械学报,2013,44(3):197-203Feng J,Liu G,Wang S W,Ma X D,Zhou W.Multi-source images registration for harvesting robot to recognize fruits[J].Transactions of the Chinese Society for Agricultural Machinery,2013,44(3):197-203(in Chinese)
[12]马晓丹,刘刚,冯娟,周薇.成熟期苹果树冠层器官异源图像配准[J].农业机械学报,2014,45(4):82-88,140Ma X D,Liu G,Feng J,Zhou W. Multi-source image registration for canopy organ of apple trees in mature period[J].Transactions of the Chinese Society for Agricultural Machinery,2014,45(4):82-88,140(in Chinese)
[13] Wang F B,Tu P,Wu C,Chen L,Feng D.Multi-image mosaic with SIFT and vision measurement for microscale structures processed by femtosecond laser[J].Optics and Lasers in Engineering,2018,100:124-130
[14] Pan L,Shi F,Zhu W,Guan L,Chen X.Detection and registration of vessels for longitudinal 3Dretinal OCT images using SURF[C].In:Medical Imaging 2018:Biomedical Applications in Molecular,Structural,and Functional Imaging.Houston:SPIE,2018,10578:105782
[15]毛雁明,兰美辉,王运琼,冯乔生.一种改进的基于Harris的角点检测方法[J].计算机技术与发展,2009,19(5):130-133Mao Y M,Lan M H,Wang Y Q,Feng Q S.An improved corner detection method based on harris[J].Computer Technology and Development,2009,19(5):130-133(in Chinese)
[16] Byravan A,Fox D.Se3-nets:Learning rigid body motion using deep neural networks[C].In:IEEE International Conference on Robotics and Automation(ICRA)2017.Singapore:IEEE,2017:173-180
[17] Efimov A I,Novikov A I.An algorithm for multistage projective transformation adjustment for image superimposition[J].Computer Optics,2016,40(2):258-265
[18] Kuroki Y,Takenaka K,Motomatsu S.Robust keypoint detection against affine transformation using moment invariants on intrinsic mode function[C].In:International Conference on Intelligent Informatics and Biomedical Sciences(ICIIBMS)2015.Okinawa:IEEE,2015:403-407
[19] Zhong Z,Li J,Luo ZM,Chapman M.Spectral-spatial residual network for hyperspectral image classification:a 3-D deep learning framework[J].IEEE Transactions on Geoscience and Remote Sensing,2018,56(2):847-858
[20] Du H,Xu Z,Ding Y.The fast lane detection of road using RANSAC algorithm[C].In:International Conference on Applications and Techniques in Cyber Security and Intelligence.Ningbo:Edizioni della Normale,2017:1-7
[21]黄梅.基于改进RANSAC算法的图像拼接技术[J].海南大学学报:自然科学版,2011,29(2):172-177Huang M.Image mosaic method based on improvedRANSAC[J].Journal of HaiNan University:Natural Science,2011,29(2):172-177(in Chinese)
[22]吴恩生,朱敏琛.一种融合局部与全局信息的距离约束角点匹配算法[J].计算机应用,2010,30(1):68-70,81Wu E S,Zhu M C.Corner matching method of constraints of distance combining local and global information[J].Journal of Computer Applications,2010,30(1):68-70,81(in Chinese)