您的位置:山东大学 -> 科技期刊社 -> 《山东大学学报(工学版)》

山东大学学报 (工学版) ›› 2019, Vol. 49 ›› Issue (2): 96-101.doi: 10.6040/j.issn.1672-3961.0.2018.242

• 机器学习与数据挖掘 • 上一篇    下一篇

基于多方法融合的非监督彩色图像分割

董新宇1,2,3(),陈瀚阅1,2,*(),李家国3,孟庆岩3,邢世和1,2,张黎明1,2   

  1. 1. 福建农林大学 资源与环境学院,福建 福州 350002
    2. 土壤生态系统健康与调控福建省高等学校重点实验室,福建 福州 350002
    3. 中国科学院遥感与数字地球研究所,北京 100101
  • 收稿日期:2018-06-07 出版日期:2019-04-20 发布日期:2019-04-19
  • 通讯作者: 陈瀚阅 E-mail:dsammy@126.com;chenhanyue.420@163.com
  • 作者简介:董新宇(1994—),男,河南商丘人,硕士研究生,主要研究方向为遥感图像处理.E-mail: dsammy@126.com
  • 基金资助:
    海南自然科学基金创新研究团队资助项目(2017CXTD015);高分辨率对地观测系统重大专项资助项目(30-Y20A07-9003-17/18);国家自然科学基金资助项目(41401399)

An unsupervised color image segmentation method based on fusion of multiple methods

Xinyu DONG1,2,3(),Hanyue CHEN1,2,*(),Jiaguo LI3,Qingyan MENG3,Shihe XING1,2,Liming ZHANG1,2   

  1. 1. College of Resources and Environment, Fujian Agriculture and Forestry University, Fuzhou 350002, Fujian, China
    2. University Key Lab of Soil Ecosystem Health and Regulation in Fujian, Fuzhou 350002, Fujian, China
    3. The Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences, Beijing 100101, Beijing, China
  • Received:2018-06-07 Online:2019-04-20 Published:2019-04-19
  • Contact: Hanyue CHEN E-mail:dsammy@126.com;chenhanyue.420@163.com
  • Supported by:
    海南自然科学基金创新研究团队资助项目(2017CXTD015);高分辨率对地观测系统重大专项资助项目(30-Y20A07-9003-17/18);国家自然科学基金资助项目(41401399)

摘要:

针对传统K-means聚类彩色图像分割方法需要人为设定初始分割类别数目、易受噪声干扰等缺陷,提出一种多方法融合非监督彩色图像分割算法。该算法对原始图像进行光谱信息增强处理以提高图像信息提取效率,对K-means聚类引入戴维森堡丁指数(Davies-Bouldin index, DBI)自动化确定最佳分割类别数目,通过图像聚类分析并进行像素标签标记,并结合高斯马尔科夫随机场(Gauss-Markov random field, GMRF)理论对标记图像进行分割,最后使用形态学算子进行后处理完成分割操作。试验结果表明。本研究方法具有一定的鲁棒性,且分割效果更接近真实性。通过对分割结果进行量化评价,进一步说明本研究方法在分割精度和准确性方面更具优势。

关键词: 彩色图像分割, 去相关拉伸, K-means聚类, 高斯马尔科夫随机场, 数学形态学算子

Abstract:

An unsupervised color image segmentation method based on fusion of multiple methods was proposed, which considered the defects of traditional K-means clustering color image segmentation method, such as the need to set the number of initial segmentation categories artificially and the vulnerability to noise interference, etc. First of all, the original image was processed by spectral information enhancement to improving the efficiency of image information extraction. Next, the number of K-means clustering segmentation categories was determined automatically by using Davies-Bouldin Index, and the clustering analysis was carried out for images and each pixel in an image was labeled. Then, the labeled image was segmented by combining the Gauss-Markov random field theory. Finally, the image after-processing was made based on the morphological operators. The segmentation experiments were carried out by using different methods, the results showed that the segmentation effect of the proposed method was closer to the origin image, and the proposed method had good robustness. And the results of quantitative evaluation of segmentation showed that this method had more advantages in segmentation precision and accuracy.

Key words: color image segmentation, decorrelation stretch, K-means clustering, Gauss-Markov random field, morphological operators

中图分类号: 

  • TP751

图1

原始影像"

图2

去相关拉伸处理结果"

表1

去相关拉伸处理前后波段的相关系数"

阶段 波段 Band1 Band2 Band3
Band1 1.000 0 0.976 1 0.772 0
处理前 Band2 0.976 1 1.000 0 0.884 5
Band3 0.772 0 0.884 5 1.000 0
Band1 1.000 0 0.308 7 0.002 9
处理后 Band2 0.308 7 1.000 0 0.039 3
Band3 0.002 9 0.039 3 1.000 0

图3

12003号图片分割结果对比"

图4

124084号图片分割结果对比"

图5

24036号图片分割结果对比"

表2

Dice系数对比"

编号 本研究算法 K-means FCM FRFCM
12003 0.968 8 0.880 3 0.796 6 0.787 6
124084 0.960 8 0.831 4 0.434 7 0.909 6
24036 0.982 6 0.978 6 0.319 3 0.979 6

表3

Jaccard相似性系数对比"

编号 本研究算法 K-means FCM FRFCM
12003 0.939 4 0.786 1 0.662 0 0.649 6
124084 0.934 1 0.711 5 0.277 7 0.834 1
24036 0.965 8 0.958 1 0.190 0 0.959 9
1 宋淑娜.基于自适应背景区域选择的图像分割方法[D].江苏科技大学, 2010.
SONG Shuna. Image segmentation methods based on adaptive background region choice[J]. Jiangsu University of Science and Technology, 2010.
2 INCETAS MO , DEMIRCI R , YAVUZCAN HG . Automatic segmentation of color images with transitive closure[J]. AEU-Int J Electron C, 2014, 68 (3): 260- 269.
doi: 10.1016/j.aeue.2013.09.001
3 陈坤, 马燕, 李顺宝. 融合直方图阈值和K-均值的彩色图像分割方法[J]. 计算机工程与应用, 2013, 49 (4): 170- 173.
doi: 10.3778/j.issn.1002-8331.1108-0301
CHEN Kun , MA Yan , LI Shunbao . Color image segmentation method by using histogram threshold and K-means[J]. Computer Engineering and Applications, 2013, 49 (4): 170- 173.
doi: 10.3778/j.issn.1002-8331.1108-0301
4 杨家红, 刘杰, 钟坚成, 等. 结合分水岭与自动种子区域生长的彩色图像分割算法[J]. 中国图象图形学报, 2010, 15 (1): 63- 68.
YANG Jiahong , LIU Jie , ZHONG Jiancheng , et al. A color image segmentation algorithm by integrating watershed with automatic seeded region growing[J]. Journal of Image and Graphics, 2010, 15 (1): 63- 68.
5 刘战杰, 马儒宁, 邹国平, 等. 一种新的基于区域生长的彩色图像分割算法[J]. 山东大学学报(理学版), 2010, 45 (7): 76- 80.
LIU Zhanjie , MA Runing , ZOU Guopin , et al. An algorithm for color image segmentation based on region growth[J]. Journal of Shandong University (Natural Science), 2010, 45 (7): 76- 80.
6 王江涛, 练煜, 石红岩. 结合模糊C均值聚类和边缘检测算法的彩色图像分割[J]. 兰州文理学院学报(自然科学版), 2015, 29 (2): 61- 65.
doi: 10.3969/j.issn.2095-6991.2015.02.015
WANG Jiangtao , LIAN Yu , SHI Hongyan . New color image segmentation based on C means clustering and edge detection algorithm[J]. Journal of Lanzhou University of Arts and Science(Nature Sciences), 2015, 29 (2): 61- 65.
doi: 10.3969/j.issn.2095-6991.2015.02.015
7 王学松, 周明全, 樊亚春, 等. 彩色图像色度距离权值的图论分割算法[J]. 中国图象图形学报, 2011, 16 (2): 221- 226.
WANG Xuesong , ZHOU Mingquan , FAN Yachun , et al. The algorithm of graph cut using HIS weights in color image segmentation[J]. Journal of Image and Graphics, 2011, 16 (2): 221- 226.
8 李震, 洪添胜, 曾祥业, 等. 基于K-means聚类的柑橘红蜘蛛图像目标识别[J]. 农业工程学报, 2012, 28 (23): 147- 153.
LI Zhen , HONG Tiansheng , ZENG Xiangye , et al. Citrus red mite image target identification based on K-means clustering[J]. Transactions of the Chinese Society of Agricultural Engineering, 2012, 28 (23): 147- 153.
9 常竞, 王玲. 一种基于遗传算法的彩色图像分割改进算法[J]. 计算机应用与软件, 2008, (8): 221- 223.
doi: 10.3969/j.issn.1000-386X.2008.08.081
CHANG Jing , WANG Ling . An improved method of color image segmentation based on genetic algorithm[J]. Computer Applications and Software, 2008, (8): 221- 223.
doi: 10.3969/j.issn.1000-386X.2008.08.081
10 李冠林, 马占鸿, 黄冲, 等. 基于K-means硬聚类算法的葡萄病害彩色图像分割方法[J]. 农业工程学报, 2010, 26 (增刊2): 32- 37.
LI Guanlin , MA Zhanhong , HUANG Chong , et al. Segmentation of color image of grape diseases using K-means clustering algorithm[J]. Transactions of the Chinese Society of Agricultural Engineering, 2010, 26 (Suppl.2): 32- 37.
11 DUbeyubey S R , DIXIT P , SINGH N , et al. Infected fruit part detection using K-means clustering segmentation technique[J]. International Journal of Interactive Multimedia and Artificial Intelligence, 2013, (2): 65- 72.
12 苗玉彬, 翟鹏. 基于改进FCM的葡萄根系图像分割算法[J]. 中国农机化学报, 2014, 35 (6): 69- 71.
MIAO Yubin , ZHAI Peng . Grape root image segmentation algorithm based on improved FCM[J]. Journal of Chinese Agricultural Mechanization, 2014, 35 (6): 69- 71.
13 胡学刚, 段瑶. 基于FCM聚类的自适应彩色图像分割算法[J]. 计算机工程与设计, 2018, 39 (7): 1984- 1989.
HU Xuegang , DUAN Yao . Adaptive color image segmentation algorithm based on FCM clustering[J]. Computer Engineering and Design, 2018, 39 (7): 1984- 1989.
14 LEI Tao , JIA Xiaohong , ZHANG Yanning , et al. Significantly fast and robust fuzzy c-means clustering algorithm based on morphological reconstruction and membership filtering[J]. IEEE Transactions on Fuzzy Systems, 2018, 26 (5): 3027- 3041.
doi: 10.1109/TFUZZ.2018.2796074
15 邓睿, 黄敬峰, 王福民. 基于去相关拉伸光谱增强的HJ-1影像水体信息提取方法研究[J]. 光谱学与光谱分析, 2011, 31 (11): 3064- 3068.
doi: 10.3964/j.issn.1000-0593(2011)11-3064-05
DENG Rui , HUANG Jingfeng , WANG Fumin . Research on extraction method of water body with DS spectral enhancement based on HJ-1 images[J]. Spectroscopy and Spectral Analysis, 2011, 31 (11): 3064- 3068.
doi: 10.3964/j.issn.1000-0593(2011)11-3064-05
16 冯超.K-means聚类算法的研究[D].大连理工大学, 2007.
FENG Chao. Research of K-means clustering algorithm[D]. Dalian: Dalian University of Technology, 2007.
17 胡钦瑞, 肖国强. 基于粗糙集和MRF的彩色图像分割方法[J]. 西南师范大学学报(自然科学版), 2014, 39 (4): 113- 119.
HU Qinrui , XIAO Guoqiang . On color image segmentation based on rough set and MRF[J]. Journal of Southwest China Normal University(Nature Science Edition), 2014, 39 (4): 113- 119.
18 詹曙, 梁植程, 谢栋栋. 前列腺磁共振图像分割的反卷积神经网络方法[J]. 中国图象图形学报, 2017, 22 (4): 516- 522.
ZHAN Shu , LIANG Zhicheng , XIE Dongdong . Deconvolutional neural network for prostate MRI segmentation[J]. Journal of Image and Graphics, 2017, 22 (4): 516- 522.
19 瞿绍军, 李乔良, 陈明, 等. 融合超像素和伪流算法的交互式图像分割[J]. 小型微型计算机系统, 2018, 39 (12): 2710- 2715.
doi: 10.3969/j.issn.1000-1220.2018.12.027
QU Shaojun , LI Qiaoliang , CHEN Ming , et al. Interactive image segmentation algorithm based on superpixel and pseudoflow[J]. Journal of Chinese Computer Systems, 2018, 39 (12): 2710- 2715.
doi: 10.3969/j.issn.1000-1220.2018.12.027
[1] 冯显英 张成梁 杨丙生 李蕾. 基于RGB颜色空间的异性纤维识别检测算法[J]. 山东大学学报(工学版), 2009, 39(5): 68-72.
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
[1] 王素玉,艾兴,赵军,李作丽,刘增文 . 高速立铣3Cr2Mo模具钢切削力建模及预测[J]. 山东大学学报(工学版), 2006, 36(1): 1 -5 .
[2] 李 侃 . 嵌入式相贯线焊接控制系统开发与实现[J]. 山东大学学报(工学版), 2008, 38(4): 37 -41 .
[3] 陈瑞,李红伟,田靖. 磁极数对径向磁轴承承载力的影响[J]. 山东大学学报(工学版), 2018, 48(2): 81 -85 .
[4] 浦剑1 ,张军平1 ,黄华2 . 超分辨率算法研究综述[J]. 山东大学学报(工学版), 2009, 39(1): 27 -32 .
[5] 刘文亮,朱维红,陈涤,张泓泉. 基于雷达图像的运动目标形态检测及跟踪技术[J]. 山东大学学报(工学版), 2010, 40(3): 31 -36 .
[6] Yue Khing Toh1 , XIAO Wendong2 , XIE Lihua1 . 基于无线传感器网络的分散目标跟踪:实际测试平台的开发应用(英文)[J]. 山东大学学报(工学版), 2009, 39(1): 50 -56 .
[7] 孙炜伟,王玉振. 考虑饱和的发电机单机无穷大系统有限增益镇定[J]. 山东大学学报(工学版), 2009, 39(1): 69 -76 .
[8] 孙殿柱,朱昌志,李延瑞 . 散乱点云边界特征快速提取算法[J]. 山东大学学报(工学版), 2009, 39(1): 84 -86 .
[9] 杨发展1 ,艾兴1 ,赵军1 ,侯建锋2 . ZrO2含量对WC基复合材料的力学性能和微观结构的影响[J]. 山东大学学报(工学版), 2009, 39(1): 92 -95 .
[10] 岳远征. 远离平衡态玻璃的弛豫[J]. 山东大学学报(工学版), 2009, 39(5): 1 -20 .