WANG Xiao-peng, ZHAO Jun-jun, MA Peng, YAO Li-juan
（School of Electronic and Information Engineering, Lanzhou Jiaotong University, Lanzhou 730070, China）
Abstract： Watershed segmentation is sensitive to noises and irregular details within the image, which frequently leads to a serious over-segmentation. Linear filtering before watershed segmentation can reduce over-segmentation to some extent, however, it often causes the position offset of object contours. For the purpose of reducing over-segmentation to preserve the location of object contours, the watershed segmentation based on the hierarchical multi-scale modification of morphological gradient is proposed. Firstly, multi-scale morphological filtering was employed to smooth the original image. Then, the gradient image was divided into multi-levels by the volume of three-dimension topographic relief, where the lower gradient layers were further modified by morphological closing with larger-sized structuring-elements, and the higher layers with the smaller one. In this way, most local minimums caused by irregular details and noises can be removed, while region contour positions corresponding to the target area were largely preserved. Finally, morphological watershed algorithm was employed to implement segmentation on the modified gradient image. The experimental results show that the proposed method can greatly reduce the over-segmentation of the watershed and avoid the position offset of the object contours.
Key words： watershed segmentation; gradient modification; hierarchical multi-scale morphological filtering; structuring element
CLD number： TN911.73 Document code： A
Article ID： 1674-8042（2017）01-0060-08doi： 10.3969/j.issn.1674-8042-2017-01-010
References
［1］Hsu W Y. Improved watershed transform for tumor segmentation: application to mammogram image compression. Expert Systems with Applications, 2012, 39(4)： 3950-3955.
［2］Siddiqui F K, Richhariya V. An efficient image segmentation approach through enhanced watershed algorithm. Computer Engineering and Intelligent Systems, 2013, 4(6)： 1-7.
［3］Thilagamani S, Shanthi N. A novel recursive clustering algorithm for image oversegmentation. European Journal of Scientific Research, 2011, 52(3)： 430-436.
［4］Vijaya K S, Lazarus, Naveen M, et al. A novel method for the detection of microcalcifications based on Multi-scale morphological gradient watershed segmentation algorithm. International Journal of Engineering Science and Technology, 2010, 2(7)： 2616-2622.
［5］Nguyen H T, Worning M. Watersnakes： energy- driven watershed segmentation. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2003, 25(3)： 330-342.
［6］Susanta M, Bhabatosh C. Multiscale morphological segmentation of gray-scale images. IEEE Transactions on Image Processing, 2004, 12(5)： 533-549.
［7］Meyer F. Levelings, image simplification filters for segmentation. Journal of Mathematical Imaging and Vision, 2004, 20(1/2)： 59-72.
［8］Leyza B D, Rodrigo M, Neucimar J L. Semiautomatic white blood cell segmentation based on multiscale analysis. IEEE Journal of Biomedical and Health Informatics, 2013, 17(1)： 250-256.
［9］Rafael V M, Jesus A, Jean S. Anisotropic morphological filters with spatially-variant structuring elements based on image-dependent gradient fields. IEEE Transactions on Image Processing, 2011, 20(1)： 200-212.
［10］Kumar S V, Lazarus M N, Nagaraju C. A novel method for the detection of microcalcifications based on multi-scale morphological gradient watershed segmentation algorithm. International Journal of Engineering Science and Technology, 2010, 2(7)： 2616-2622.
［11］Corinne V, Fernand M. The viscous watershed transform. Journal of Mathematical Imaging and Vision, 2005, 22(2)： 251-267.
［12］Kande G, Savithri T, Subbaiah P. Segmentation of vessels in fundus images using spatiallly weighted fuzzy c-means clustering algorithm. International Journal of Computer Science and Network Security, 2007,7(12)： 102-109.
［13］QIN Kun, XU Min. Remote sensing image segmentation based on cloud model and FCM. Geo-Information Science, 2008, 10(3)： 302-307.
［14］NING Ji-feng, ZHANG Lei, ZHANG David, et al. Interactive image segmentation by maximal similarity based region merging. Pattern Recognition, 2010, 43(2)： 445-456.
［15］Stelios K, Vassilios C. A robust fuzzy local information c-means clustering algorithm. IEEE Transactions on Image Processing, 2010, 19(5)： 1328-1337.
［16］CHEN Jian, YAN Bin, JIANG Hua, et al. Interactive image segmentation by improved maximal similarity based region merging. In： Proceedings of 2013 IEEE International Conference on Medical Imaging Physics and Engineering (ICMIP201E), Washington DC, 2013： 279-282.
［17］Rajendran A, Dhanasekaran R. Fuzzy clustering and deformable model for tumor segmentation on MRI brain image： a combined approach. Procedia Engineering, 2012, 30(4)： 327-333.
［18］Vovk U, Pernus F, Likar B. A review of methods for correction of intensity inhomogeneity in MRL. IEEE Transactions on Medical Imaging, 2007, 26(3)： 405-421.
［19］WANG Xiao-peng, DANG Jian-wu, WANG Yang-ping, Parameterized morphological watershed segmentation, Journal of the China Railway Society, 2013, 35(1)： 66-70.
［20］WANG Xiao-peng, HAO Chong-yang, FAN Yang-yu. Watershed segmentation based on morphological scale-space and gradient modification, Journal of Electronics & Information Technology, 2006, 28(3)： 485-489.
［21］LIU Yue, WANG Xiao-peng, WANG Jin-quan, et al. Watershed algorithm for brain tumor segmentation based on morphological reconstruction and gradient layered modification, Application Research of Computers, 2015, 32(8)： 2487-2491.
基于分层多尺度形态学梯度修正的分水岭分割
王小鹏， 赵君君， 马鹏， 姚丽娟
（兰州交通大学 电子与信息工程学院， 甘肃 兰州 730070）
摘要： 形态学分水岭分割对图像中的噪声和非规则细节较为敏感， 常常导致较严重的过分割。 如果在分水岭分割之前采用线性滤波器进行平滑，可以在某种程度上消除噪声和非规则细节干扰造成的分水岭过分割， 但是可能使分割出的目标轮廓产生位置偏移。 为了能够在消除过分割的同时保持目标轮廓的位置不变， 提出了一种基于分层多尺度形态学梯度修正的分水岭分割方法。 首先对原始图像进行多尺度形态学滤波平滑； 然后根据形态学梯度图像的三维地貌体积对其进行分层多尺度修正， 自适应地确定修正所需的结构元素尺寸， 对于低梯度层级采用较大尺寸结构元素进行形态学闭运算， 消除因非规则细节产生过分割的非规则局部极小值， 而对较高梯度层则采用较小尺寸的结构元素， 保持区域轮廓的位置不变； 最后在修正梯度图像基础上， 运用标准分水岭变换实现图像分割。 实验结果表明， 该方法能够在消除过分割的同时， 较准确的保持目标轮廓的位置。
关键词： 分水岭分割; 梯度修正; 分层多尺度形态学滤波; 结构元素
引用格式：WANG Xiao-peng, ZHAO Jun-jun, MA Peng, et al. Watershed segmentation based on hierarchical multi-scale modification of morphological gradient. Journal of Measurement Science and Instrumentation, 2017, 8(1)： 60-67. ［doi： 10.3969／j.issn.1674-8042.2017-01-010］
[full text view]
