基于CV模型改进的磁共振成像图像分割方法  

Improved Magnetic Resonance Image Segmentation Method Based on CV Model

作　　者：兰红[1] 韩纪东 LAN Hong;HAN Ji-dong(School of Information Engineering,Jiangxi University of Science and Technology,Ganzhou 34100,China)

机构地区：[1]江西理工大学信息工程学院,赣州341000

出　　处：《科学技术与工程》2018年第28期229-234,共6页Science Technology and Engineering

基　　金：国家自然科学基金(61762046);江西省教育厅科技重点项目(GJJ160599);江西省自然科学基金(20161BAB212048);江西省研究生创新专项基金(YC2017-S301)资助

摘　　要：为解决灰度变化缓慢、边缘变化不明显的磁共振成像(magnetic resonance imaging,MRI)图像的分割问题,在CV(Chan-Vese)模型的基础上,改进了CV模型的能量泛函,同时用新的边缘指示函数来替换Dirac函数,优化了CV模型的参数,提高了CV模型的分割精度和分割速度。首先,引入了一个新的局部项。用局部直方图均衡化预处理过的图像与原图像相减得到目标边缘变化较为明显的图像,并将其作为局部项引入到CV模型的能量泛函。然后,由局部项构建新的边缘指示函数。用新构建的边缘指示函数代替Dirac函数,解决了CV模型演化曲线不能检测远离目标的边缘的问题。最后,优化平滑项参数,减少迭代次数提高运行效率。实验结果显示,算法对脑部复发性胶质母细胞瘤的MRI图像具有较好的分割效果。In order to solve the problem of segmentation of magnetic resonance imaging(MRI)images with slow gray change and no obvious edge,the energy functional of Chan-Vese(CV)model was improved,replaced the Dirac function with new edge indication function,and optimized the parameters of the CV model.First,a new local term was introduced.An image is obtained by subtracting the original image from the image of the local histogram equalization,which is introduced into the energy functional of CV model as a local term.Then,a new edge indication function was constructed from the local terms.The newly constructed edge indication function was used instead of the Dirac function to solve the problem that the evolution curve of the CV model can not detect the edge far from the target.Finally,the parameters of smoothing items were optimized to reduce the number of iterations as well as improve the operational efficiency.The result of experiment shows that the proposed algorithm has a good segmentation effect on MRI images of brain recurrent glioblastoma.

关 键 词：活动轮廓模型 CV模型 MRI图像分割 局部直方图均衡化 边缘指示函数 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]