面向医学图像分割的直线截距直方图倒数交叉熵方法

发布时间：2018-01-15 16:16

本文关键词：面向医学图像分割的直线截距直方图倒数交叉熵方法　出处：《数据采集与处理》2015年05期 　论文类型：期刊论文

【摘要】：为了进一步提高医学图像分割的速度和准确度,为临床诊断和辅助治疗提供更为充分有效的依据,本文提出了一种基于直线截距直方图的倒数交叉熵图像阈值分割方法。首先定义了直线截距直方图;然后根据医学图像的二维信息,建立该图像的直线截距直方图;最后,推导出基于该直方图的倒数交叉熵阈值选取准则,并以此对医学图像进行分割。实验结果表明,与基于混沌小生境粒子群优化(Niche chaotic mutation particle swarm optimization,NCPSO)的二维倒数熵法、基于分解的二维指数灰度熵法、基于斜分的二维对称交叉熵法及基于粒子群优化(Particle swarm optimization,PSO)的二维Tsallis交叉熵法相比,本文方法分割后的图像中目标区域完整准确,边缘细节清晰丰富,且所需运行时间大幅减少,是医学影像研究中可选择的一种快速有效的图像分割方法。
[Abstract]:In order to further improve the speed and accuracy of medical image segmentation, for clinical diagnosis and auxiliary treatment to provide a more sufficient and effective basis. In this paper, a threshold segmentation method of reciprocal cross-entropy image based on straight line intercept histogram is proposed. Firstly, the straight line intercept histogram is defined. Then, according to the two-dimensional information of the medical image, the histogram of the straight line intercept of the image is established. Finally, the threshold selection criterion of reciprocal cross-entropy based on the histogram is derived, and the medical image is segmented. The experimental results show that. And the optimization of niche chaotic mutation particle swarm optimization based on chaos. NCPSO (2-D reciprocal Entropy method) and 2-D exponential Gray Entropy method based on decomposition. Two-dimensional symmetric cross-entropy method based on oblique division and particle swarm optimization based on particle swarm optimization. Compared with the two-dimensional Tsallis cross-entropy method, the proposed method has the advantages of complete and accurate target region, rich edge details, and reduced running time. It is a fast and effective image segmentation method in medical image research.
【作者单位】：南京航空航天大学电子信息工程学院;
【基金】：国家自然科学基金(60872065)资助项目江苏省普通高校研究生科研创新计划(SJLX15_0116)资助项目江苏高校优势学科建设工程(2012)资助项目中央高校基本科研业务费专项资金资助项目
【分类号】：R445;TP391.41
【正文快照】： 引言随着X射线、计算机断层扫描成像(Computed tomography,CT)、磁共振成像(Magnetic resonanceimaging,MRI)以及磁共振血管造影(Magnetic resonance angiography,MRA)等技术的不断发展,医学影像在现代临床诊断和医学研究领域中的应用价值也日益重要[1-3]。医学图像数据源丰富

【参考文献】