摘要
为提高对右心室形态结构和功能异常的检测和诊断对心脑血管疾病的作用,并针对其心腔几何形状复杂,解剖结构特殊,使心脏磁共振图像准确分割变成难点,提出通过OMP和K-SVD的有效结合方法。首先将字典学习过程和稀疏表示结合,这样能更好的训练结果,然后由这些训练出的字典模型和稀疏表示系数进行图像的特征重构,最后根据重构后的误差进行分类实现分割。以MICCAI右心室分割挑战中提供的数据为依托,按照如上方法对右心室进行分割,并对分割结果进行分析,实验结果证明提出的基于K-SVD训练字典的分割方法,分割结果较为准确,从而表明了此方法的有效性。
To improve the structure and function of right ventricular morphology of anomaly detection and diagnosis of cardiovascular and cerebrovascular diseases,and aims at its heart chamber geometry is complex,special anatomical structures,and an exact cardiac magnetic resonance image segmentation into difficulties,put forward by combining the OMP and K-SVD method. Dictionary of the learning process is combined with sparse representation,it can better training result,then the dictionary by these training model and the characteristics of the coefficient of sparse representation for image reconstruction,the division of classified according to the error of reconstruction after the implementation. With MICCAI right ventricle segmentation based on data provided by the challenge,in accordance with the above method to split,the right ventricle and the segmentation result is analyzed,the experimental results show that the proposed segmentation method based on K-SVD training dictionary,segmentation result is relatively accurate,so as to show the effectiveness of this method.
引文
[1]雷琼.基于Matlab图像分割的研究[J].电子设计工程,2015,23(21):167-169.
[2]王改云,杨小娟.改进的分水岭算法在图像分割中的研究[J].电子设计工程,2014,22(10):64-67.
[3]王雅宁,梁新刚.基于交互式图像分割的立体匹配方法[J].计算机技术与发展,2016,26(9):163-166.
[4]朱梓豪,张帅,葛欣.基于机器学习的图像边缘检测方法的研究与应用[J].信息与电脑(理论版),2018(4):56-58.
[5]刘燕,董蓉,李勃.基于机器学习的图像分割算法研究[J].电视技术,2017,41(Z4):32-39.
[6]Guo Y,Gao Y,Shao Y,et al. Deformable segmen?tation of 3D MR prostate images via distributed dis?criminative dictionary and ensemble learning[J].Medical Physics,2014(41):072303.
[7]Huang X,Dione D P,Compas C.et al. Contour tracking in echocardio graphic sequences via sparse representation and dictionary learning[J].Medical Image Analysis,2014,18(2):253-271.
[8]Zhang S,Zhan Y,Metaxas D N. Deformable seg?mentation via sparse representation and dictionary learning[J]. Medical Image Analysis,2012,16(7):1385-1396.
[9]徐伟尧.OMP算法及其在数字图像中的应用分析[J].广东通信技术,2015,35(3):71-75.
[10]石曼曼,李雷.一种改进的OMP算法及其在图像重构上的应用[J].计算机技术与发展,2018,28(2):94-97.
[11]陈成钢,艾涛.基于灰度共生矩阵的纹理分析的统计方法[J].天津农学院学报,2014,21(2):33-38.
[12]张清丽,苏士美,基于改进小波变换的QRS特征提取算法研究[J].郑州大学学报(理学版),2017,49(4):100-103.
[13]张东,唐向宏,张少鹏,等.小波变换与纹理合成相结合的图像修复[J].中国图象图报,2015,20(7):882-894.
[14]赵红丹,田喜平.基于K-means算法分割遥感图像的阈值确定方法研究[J].科学技术与程,2017,17(9):250-254.
[15]冯波,郝文宁,陈刚,等.K-means算法初始聚类中心选择的优化[J].计算机工程用,2013,49(14):182-185,192.
[16]谢娟英,王艳娥.最小方差优化初始聚类中心的K-means算法[J].计算机工程,2014,40(8):205-211,223.
[17]左宇,顾琨,李博江,等.基于改进K-均值聚类的电动机群聚合[J].陕西电力,2016(10):32-37.
[18]张彦军,张姗姗.基于模糊C均值聚类算法的海洋环境监测研究[J].工业仪表与自动化装置,2017(6):3-6.