基于二项分布的双窗OTSU的矿石分割模型
|
摘要
表面不均匀的矿石视频图像目标受光照、周围环境的影响,噪声严重,传统的阈值法不能准确有效的分割。而双窗OTSU方法计算代价过高,且双窗阈值取极值作为目标像素的最佳阈值,导致部分像素误分类。提出了基于二项分布优化的双窗OTSU算法,双窗模板大小取决于最大和最小目标尺寸。通过实验与数学理论证明方法的可行性与准确性,时间成本降低50%～80%。对于目标易分类的图像,分类精度提升6%左右;对于目标难以分类的图像,分类精度提升13%左右。算法自主确定窗口尺寸,增加其智能性,抗噪声干扰性强,降低了误分类率,时间成本大幅度降低。
Noise of uneven ore objects in images is seriously influenced by illumination and textures,causing it cannot be accurately and effectively segmented by traditional threshold method.The calculation cost of the doubleWindow OTSU method is too high,and the extreme value of the double-Window threshold is taken as the optimal threshold value of the target pixel,resulting in partial pixel misclassification.In this paper,a double-window OTSU algorithm based on binomial distribution optimization was presented,the size of the double window template depended on the maximum and minimum target sizes.The feasibility and accuracy of the proposed method were proved by experiment and mathematical theory,and the time cost was reduced by 50% ～80% greatly.The classification accuracy was improved by about 6% for the image of object easy classification,about 13% for the image of object uneasy classification.The algorithm determined the size of the window independently,increased its intelligence and strong anti-noise,reduced the misclassification rate and greatly reduced the time cost.
Noise of uneven ore objects in images is seriously influenced by illumination and textures,causing it cannot be accurately and effectively segmented by traditional threshold method.The calculation cost of the doubleWindow OTSU method is too high,and the extreme value of the double-Window threshold is taken as the optimal threshold value of the target pixel,resulting in partial pixel misclassification.In this paper,a double-window OTSU algorithm based on binomial distribution optimization was presented,the size of the double window template depended on the maximum and minimum target sizes.The feasibility and accuracy of the proposed method were proved by experiment and mathematical theory,and the time cost was reduced by 50% ～80% greatly.The classification accuracy was improved by about 6% for the image of object easy classification,about 13% for the image of object uneasy classification.The algorithm determined the size of the window independently,increased its intelligence and strong anti-noise,reduced the misclassification rate and greatly reduced the time cost.
引文
[1]陈炳权,刘宏立,孟凡斌.数字图像处理技术的现状及其发展方向[J].吉首大学学报(自然科学版),2009,30(1):63-70.
[2]徐彩云.图像分割算法的研究综述[J].电脑知识与技术,2014,10(11):2637-2639.
[3]罗希平,田捷,诸葛婴,等.图像分割方法综述[J].模式识别与人工智能,1999(3):300-312.
[4]SEZGIN M,SANKUR B.Survey over image thresholding techniques and quantitative performance evaluation[J].JElectronic Imaging,2004,13(1):146-168.
[5]ZHANG G Y,LIU G Z.Ore image thresholding using bi-neighbourhood OTSU’s approach[J].IET Journals,2010,46(25):1666-1668.
[6]OHTSU N.A threshold selection method from gray-level histograms[J].IEEE Transactions on Systems Man&Cybernetics,1979,9(1):62-66.
[7]REDDI S S,RUDIN S F,KESHAVAN H R.An optimal multiple threshold scheme for image segmentation[J].Systems Man&Cybernetics IEEE Transactions on,1984,SMC-14(4):661-665.
[8]刘健庄,栗文青.灰度图象的二维OTSU自动阈值分割法[J].自动化学报,1993,19(1):101-105.
[9]范九伦,赵凤.灰度图像的二维OTSU曲线阈值分割法[J].电子学报,2007,35(4):751-755.
[10]ZHANG J,HU J.Image Segmentation Based on 2DOTSUMethod with Histogram Analysis[C]//International Conference on Computer Science and Software Engineering.IEEE,2008:105-108.
[11]KAUT H,SINGH R.A review on image segmentation techniques[J].Pattern Recognit,1993,26(9):1277-1294.
[12]TESSIER J,DUCHESNE C,and BARTOLACCI G.A machine vision approach to on-line estimation of run-of-mine ore composition on conveyor belts[J].Min.Eng.,2007(20):1129.
[13]KAPUR J N,SAHOO P K,WONG A K C.A new method for gray-level picture thresholding using the entropy of the histogram[J].Computer Vision,Graphics,and Image Processing,1985,29(3):273-285.
[14]SIMPHIWE M.A machine vision-based approach to measuring the size distribution of rocks on a conveyor belt[D].University of Cape Town,Cape Town,2004:23.
[15]张国英,邱波,刘冠洲.基于双窗的自适应一维熵矿石图像分割方法[C]//信息技术、服务科学与工程管理国际学术会议论文集,北京,2011.
[16]程述立,汪烈军,秦继伟,等.群智能算法优化的结合熵的最大类间方差法与脉冲耦合神经网络融合的图像分割算法[J].计算机应用,2017,37(12):3528-3535,3553.
[17]方挺,张亚萍.一种基于标识的矿石图像分割方法[J].软件导刊,2016,15(6):215-217.
[18]马连铭,姬建钢,杜波,等.基于机器视觉的矿石粒度检测方法比较研究[J].矿山机械,2017,45(10):53-59.
[19]郭庆昌,何国宝.双窗目标特征提取算法研究[J].计算机仿真,2009,26(12):177-181.
[20]CHEN Zhikun,LIU Donghai.Research on two methods of improving the threshold de-noising[C]//2010 4th International Conference on Intelligent Information Technology Application.2010,2:212-214.
[21]汪荣贵,吴昊,方帅,等.一种新的自适应二维OTSU图像分割算法研究[J].中国科学技术大学学报,2010,40(8):841-847.
[22]REN Shijun,WANG Yadong.A proof of the convergence theorem of maximum-entropy clustering algorithm[J].Science China(Information Sciences),2010,53(6):1151-1158.
[23]BRINK A D.Thresholding of digital images using two-dimensional entropies[J].Pattern Recognition,1992,25(8):803-808.
[24]DU Feng,SHI Wenkang,ZHU Zhenfu,et al.Infrared image segmentation with 2-D maximum entropy method based on particle swarm optimization(PSO)[J].Pattern Recognition Letters,2005,26(5):597-603.
[25]张雪媛.二项分布及其应用的历史研究[D].天津:天津财经大学,2015.
[2]徐彩云.图像分割算法的研究综述[J].电脑知识与技术,2014,10(11):2637-2639.
[3]罗希平,田捷,诸葛婴,等.图像分割方法综述[J].模式识别与人工智能,1999(3):300-312.
[4]SEZGIN M,SANKUR B.Survey over image thresholding techniques and quantitative performance evaluation[J].JElectronic Imaging,2004,13(1):146-168.
[5]ZHANG G Y,LIU G Z.Ore image thresholding using bi-neighbourhood OTSU’s approach[J].IET Journals,2010,46(25):1666-1668.
[6]OHTSU N.A threshold selection method from gray-level histograms[J].IEEE Transactions on Systems Man&Cybernetics,1979,9(1):62-66.
[7]REDDI S S,RUDIN S F,KESHAVAN H R.An optimal multiple threshold scheme for image segmentation[J].Systems Man&Cybernetics IEEE Transactions on,1984,SMC-14(4):661-665.
[8]刘健庄,栗文青.灰度图象的二维OTSU自动阈值分割法[J].自动化学报,1993,19(1):101-105.
[9]范九伦,赵凤.灰度图像的二维OTSU曲线阈值分割法[J].电子学报,2007,35(4):751-755.
[10]ZHANG J,HU J.Image Segmentation Based on 2DOTSUMethod with Histogram Analysis[C]//International Conference on Computer Science and Software Engineering.IEEE,2008:105-108.
[11]KAUT H,SINGH R.A review on image segmentation techniques[J].Pattern Recognit,1993,26(9):1277-1294.
[12]TESSIER J,DUCHESNE C,and BARTOLACCI G.A machine vision approach to on-line estimation of run-of-mine ore composition on conveyor belts[J].Min.Eng.,2007(20):1129.
[13]KAPUR J N,SAHOO P K,WONG A K C.A new method for gray-level picture thresholding using the entropy of the histogram[J].Computer Vision,Graphics,and Image Processing,1985,29(3):273-285.
[14]SIMPHIWE M.A machine vision-based approach to measuring the size distribution of rocks on a conveyor belt[D].University of Cape Town,Cape Town,2004:23.
[15]张国英,邱波,刘冠洲.基于双窗的自适应一维熵矿石图像分割方法[C]//信息技术、服务科学与工程管理国际学术会议论文集,北京,2011.
[16]程述立,汪烈军,秦继伟,等.群智能算法优化的结合熵的最大类间方差法与脉冲耦合神经网络融合的图像分割算法[J].计算机应用,2017,37(12):3528-3535,3553.
[17]方挺,张亚萍.一种基于标识的矿石图像分割方法[J].软件导刊,2016,15(6):215-217.
[18]马连铭,姬建钢,杜波,等.基于机器视觉的矿石粒度检测方法比较研究[J].矿山机械,2017,45(10):53-59.
[19]郭庆昌,何国宝.双窗目标特征提取算法研究[J].计算机仿真,2009,26(12):177-181.
[20]CHEN Zhikun,LIU Donghai.Research on two methods of improving the threshold de-noising[C]//2010 4th International Conference on Intelligent Information Technology Application.2010,2:212-214.
[21]汪荣贵,吴昊,方帅,等.一种新的自适应二维OTSU图像分割算法研究[J].中国科学技术大学学报,2010,40(8):841-847.
[22]REN Shijun,WANG Yadong.A proof of the convergence theorem of maximum-entropy clustering algorithm[J].Science China(Information Sciences),2010,53(6):1151-1158.
[23]BRINK A D.Thresholding of digital images using two-dimensional entropies[J].Pattern Recognition,1992,25(8):803-808.
[24]DU Feng,SHI Wenkang,ZHU Zhenfu,et al.Infrared image segmentation with 2-D maximum entropy method based on particle swarm optimization(PSO)[J].Pattern Recognition Letters,2005,26(5):597-603.
[25]张雪媛.二项分布及其应用的历史研究[D].天津:天津财经大学,2015.