基于切比雪夫不等式的紫色土彩色图像分割
|
摘要
野外采集的机器视觉图像往往包含复杂背景,会对机器视觉识别紫色土产生影响,为了避免背景干扰,分割提取紫色土区域图像是首要的工作.本文应用3×3小子阵的标准差测度,建立模型优化紫色土区域的土壤与杂质类间和类内方差比,获得优化的置信概率P和H域分割阈值,提出了一种基于切比雪夫不等式的自适应H阈值分割算法,实现基于图像自身紫色土特征的自适应分割,提升初分割出紫色土区域图像的精度.针对初分割结果中的孤立点、离散小土块和空洞,提出了从图像中心点出发的剔除背景区域孤立点和离散小土块的螺旋生长算法和基于4方向边界点确认的紫色土区域的空洞填充算法.仿真实验结果显示:自适应切比雪夫阈值分割算法与螺旋生长算法和空洞填充算法结合,分割提取出紫色土区域图像的误分率降低到3.24%,总时间花销更少,算法是有效的.
The vision images of the types of purple soils picked up in the field with an identification system based on machine vision inevitably contain the complex background. To avoid background interference, segmentation extraction of the images of the purple soil area is essential. In this paper, an optimization model is established by using standard deviation measure of the 3×3 small subarray, which minimizes the ratio of inter-class variance to intra-class variance between soil and impurities in the purple soil area. The optimized confidence probability P and the segment threshold of H domain are obtained, and an adaptive segmentation algorithm based on the Chebyshev inequality is developed by the optimization model. The algorithm realizes the adaptive segmentation based on the purple soil features in the image and improves the accuracy of preliminary segmenting of the image of purple soil area. Then, a spiral growth algorithm and a hole-filling algorithm are proposed to process the raw segmentation result. They remove the isolated points and small soil block from the background area and fill in the hole in the purple soil area to extract the image of the purple soil area from the vision image. The simulation results show that the error rate of the proposed method is reduced to 3.24% and the total time spent is less. The proposed adaptive Chebyshev threshold segmentation algorithm combined with spiral growth algorithm and cavity filling algorithm can effectively extract purple soil region image.
The vision images of the types of purple soils picked up in the field with an identification system based on machine vision inevitably contain the complex background. To avoid background interference, segmentation extraction of the images of the purple soil area is essential. In this paper, an optimization model is established by using standard deviation measure of the 3×3 small subarray, which minimizes the ratio of inter-class variance to intra-class variance between soil and impurities in the purple soil area. The optimized confidence probability P and the segment threshold of H domain are obtained, and an adaptive segmentation algorithm based on the Chebyshev inequality is developed by the optimization model. The algorithm realizes the adaptive segmentation based on the purple soil features in the image and improves the accuracy of preliminary segmenting of the image of purple soil area. Then, a spiral growth algorithm and a hole-filling algorithm are proposed to process the raw segmentation result. They remove the isolated points and small soil block from the background area and fill in the hole in the purple soil area to extract the image of the purple soil area from the vision image. The simulation results show that the error rate of the proposed method is reduced to 3.24% and the total time spent is less. The proposed adaptive Chebyshev threshold segmentation algorithm combined with spiral growth algorithm and cavity filling algorithm can effectively extract purple soil region image.
引文
[1] 杨转.基于HSI颜色模型的杂草与土壤背景分割方法研究 [J].河北农业大学学报,2011,34(4):124-127.
[2] 刘絮雨,张相芬,马燕,等.基于改进空间模糊聚类的DTI图像分割算法 [J].中国生物医学工程学报,2018,37(4):394-403.
[3] 徐黎明,吕继东.基于同态滤波和K均值聚类算法的杨梅图像分割 [J].农业工程学报,2015,31(14):202-208.
[4] 张玲,余攀,刘艺.具强噪声鲁棒性的聚类图像分割算法 [J].西南师范大学学报(自然科学版),2016,41(2):99-106.
[5] 刁智华,刁春迎,袁万宾,等.基于改进型模糊边缘检测的小麦病斑阈值分割算法 [J].农业工程学报,2018,34(10):147-152.
[6] 刘向南,王昱潭,赵琛,等.基于分水岭算法的灵武长枣图像分割方法研究 [J].计算机工程与应用,2018,54(15):169-175,184.
[7] 王万耀,段先华,徐丹,等.基于显著性的Grabcut图像分割方法 [J].计算机工程,2018,44(7):230-236,243.
[8] 王见,周勤,尹爱军.改进Otsu算法与ELM融合的自然场景棉桃自适应分割方法 [J].农业工程学报,2018,34(14):173-180.
[9] 施成湘.基于邻域特征的彩色图像分割方法 [J].西南师范大学学报(自然科学版),2015,40(2):106-110.
[10] SZIRANYI T,SHADAYDEH M.Segmentation of Remote Sensing Images Using Similarity-Measure-Based Fusion-MRF Model [J].IEEE Geoscience and Remote Sensing Letters,2014,11(9):1544-1548.
[11] HOU B,YANG C,REN B,et al.Decomposition-Feature-Iterative-Clustering-Based Superpixel Segmentation for PolSAR Image Classification [J].IEEE Geoscience and Remote Sensing Letters,2018,15(8):1239-1243.
[12] DESPOTOVIC I,VANSTEENKISTE E,PHILIPS W.Spatially Coherent Fuzzy Clustering for Accurate and Noise-Robust Image Segmentation [J].IEEE Signal Processing Letters,2013,20(4):295-298.
[13] YOSHINARI K,MURAHIRA K,HOSHI Y,et al.Color Image Enhancement in Improved HSI Color Space [C] //2013 International Symposium on Intelligent Signal Processing and Communication Systems,Naha-shi,Japan:IEEE,2013:429-434.
[14] ALI H,RADA L,BADSHAH N.Image Segmentation for Intensity Inhomogeneity in Presence of High Noise [J].IEEE Transactions on Image Processing,2018,27(8):3729-3738.
[15] HELL B,KASSUBECK M,BAUSZAT P,et al.An Approach Toward Fast Gradient-Based Image Segmentation [J].IEEE Transactions on Image Processing,2015,24(9):2633-2645.
[16] WANG M,DONG Z P,CHENG Y F,et al.Optimal Segmentation of High-Resolution Remote Sensing Image by Combining Superpixels with the Minimum Spanning Tree [J].IEEE Transactions on Geoscience and Remote Sensing,2018,56(1):228-238.
[17] MICHEL J,YOUSSEFI D,GRIZONNET M.Stable Mean-Shift Algorithm and Its Application to the Segmentation of Arbitrarily Large Remote Sensing Images [J].IEEE Transactions on Geoscience and Remote Sensing,2015,53(2):952-964.
[18] 王小平,李柳柏.基于AdaBoost算法的图像自动标注 [J].西南大学学报(自然科学版),2015,37(7):174-180.
[19] 王小娟,李云伍,刘得雄,等.基于机器视觉的丘陵山区田间道路虚拟中线提取方法 [J].西南大学学报(自然科学版),2018,40(4):162-169.
[20] 重庆市质量技术监督局.DB50/T 796-2017:重庆土壤分类与代码 [S].重庆:重庆市质量技术监督局,2017.
[21] BAI X B,LI X X,FU Z T,et al.A Fuzzy Clustering Segmentation Method Based on Neighborhood Grayscale Information for Defining Cucumber Leaf Spot Disease Images [J].Computers and Electronics in Agriculture,2017,136:157-165.
[22] RAJABY E,AHADI S M,AGHAEINIA H.Robust Color Image Segmentation Using Fuzzy C-Means with Weighted Hue and Intensity [J].Digital Signal Processing,2016,51:170-183.
[2] 刘絮雨,张相芬,马燕,等.基于改进空间模糊聚类的DTI图像分割算法 [J].中国生物医学工程学报,2018,37(4):394-403.
[3] 徐黎明,吕继东.基于同态滤波和K均值聚类算法的杨梅图像分割 [J].农业工程学报,2015,31(14):202-208.
[4] 张玲,余攀,刘艺.具强噪声鲁棒性的聚类图像分割算法 [J].西南师范大学学报(自然科学版),2016,41(2):99-106.
[5] 刁智华,刁春迎,袁万宾,等.基于改进型模糊边缘检测的小麦病斑阈值分割算法 [J].农业工程学报,2018,34(10):147-152.
[6] 刘向南,王昱潭,赵琛,等.基于分水岭算法的灵武长枣图像分割方法研究 [J].计算机工程与应用,2018,54(15):169-175,184.
[7] 王万耀,段先华,徐丹,等.基于显著性的Grabcut图像分割方法 [J].计算机工程,2018,44(7):230-236,243.
[8] 王见,周勤,尹爱军.改进Otsu算法与ELM融合的自然场景棉桃自适应分割方法 [J].农业工程学报,2018,34(14):173-180.
[9] 施成湘.基于邻域特征的彩色图像分割方法 [J].西南师范大学学报(自然科学版),2015,40(2):106-110.
[10] SZIRANYI T,SHADAYDEH M.Segmentation of Remote Sensing Images Using Similarity-Measure-Based Fusion-MRF Model [J].IEEE Geoscience and Remote Sensing Letters,2014,11(9):1544-1548.
[11] HOU B,YANG C,REN B,et al.Decomposition-Feature-Iterative-Clustering-Based Superpixel Segmentation for PolSAR Image Classification [J].IEEE Geoscience and Remote Sensing Letters,2018,15(8):1239-1243.
[12] DESPOTOVIC I,VANSTEENKISTE E,PHILIPS W.Spatially Coherent Fuzzy Clustering for Accurate and Noise-Robust Image Segmentation [J].IEEE Signal Processing Letters,2013,20(4):295-298.
[13] YOSHINARI K,MURAHIRA K,HOSHI Y,et al.Color Image Enhancement in Improved HSI Color Space [C] //2013 International Symposium on Intelligent Signal Processing and Communication Systems,Naha-shi,Japan:IEEE,2013:429-434.
[14] ALI H,RADA L,BADSHAH N.Image Segmentation for Intensity Inhomogeneity in Presence of High Noise [J].IEEE Transactions on Image Processing,2018,27(8):3729-3738.
[15] HELL B,KASSUBECK M,BAUSZAT P,et al.An Approach Toward Fast Gradient-Based Image Segmentation [J].IEEE Transactions on Image Processing,2015,24(9):2633-2645.
[16] WANG M,DONG Z P,CHENG Y F,et al.Optimal Segmentation of High-Resolution Remote Sensing Image by Combining Superpixels with the Minimum Spanning Tree [J].IEEE Transactions on Geoscience and Remote Sensing,2018,56(1):228-238.
[17] MICHEL J,YOUSSEFI D,GRIZONNET M.Stable Mean-Shift Algorithm and Its Application to the Segmentation of Arbitrarily Large Remote Sensing Images [J].IEEE Transactions on Geoscience and Remote Sensing,2015,53(2):952-964.
[18] 王小平,李柳柏.基于AdaBoost算法的图像自动标注 [J].西南大学学报(自然科学版),2015,37(7):174-180.
[19] 王小娟,李云伍,刘得雄,等.基于机器视觉的丘陵山区田间道路虚拟中线提取方法 [J].西南大学学报(自然科学版),2018,40(4):162-169.
[20] 重庆市质量技术监督局.DB50/T 796-2017:重庆土壤分类与代码 [S].重庆:重庆市质量技术监督局,2017.
[21] BAI X B,LI X X,FU Z T,et al.A Fuzzy Clustering Segmentation Method Based on Neighborhood Grayscale Information for Defining Cucumber Leaf Spot Disease Images [J].Computers and Electronics in Agriculture,2017,136:157-165.
[22] RAJABY E,AHADI S M,AGHAEINIA H.Robust Color Image Segmentation Using Fuzzy C-Means with Weighted Hue and Intensity [J].Digital Signal Processing,2016,51:170-183.