基于高斯分解的多尺度3D Otsu阈值分割算法  

Multi-scale 3D Otsu thresholding algorithm based on Gaussian decomposition

作　　者：肖明尧[1,2] 李雄飞[2] XIAO Ming-yao LI Xiong-fei(College of Computer Science and Technology, Changchun Normal University, Changchun 130032, China College of Computer Science and Technology, Jilin University, Changchun 130012, China)

机构地区：[1]长春师范大学计算机科学与技术学院,长春130032 [2]吉林大学计算机科学与技术学院,长春130012

出　　处：《吉林大学学报（工学版）》2017年第1期255-261,共7页Journal of Jilin University:Engineering and Technology Edition

基　　金：国家自然科学基金项目(61272209);国家科技支撑计划项目(2012BAH48F02)

摘　　要：针对阈值图像分割算法对噪音敏感的问题,提出了一种新的基于分解的Otsu阈值分割算法。整个分割算法为一个迭代过程,在每次迭代中,该图像首先用3D Otsu算法进行分割,然后利用高斯核函数对原图像进行滤波,得到一个平滑的图像,然后被输入到下一个迭代中。最后,合并每次迭代过程中产生的分割结果,获得最终的分割结果。该算法的优点在于分割结果稳定,且具有较强的抗噪性。本文在MR大脑图像上进行实验,结果表明,该算法优于其他同类阈值分割算法。Current thresholding algorithms for image segmentation are sensitive to noise.To overcome this problem,a new Otsu thresholding algorithm is proposed based on image decomposition.The whole segmentation algorithm is designed as an iteration procedure.In each iteration the image is segmented by the 3D Ostu,and then it is filtered by Gaussian kernel filtering to get a smoothed image,which is taken as the input of the next iteration.Finally,segmentation results obtained in the iterations and are pooled to get final segmentation.The advantages of the proposed algorithm are that its segmentation results are stable and it is robust to noise.Experiments on medical MR brain images are conducted to demonstrate the effectiveness of the proposed method.Results indicate that the proposed algorithm is superior to other thresholding algorithms.

关 键 词：计算机应用 图像分割 OTSU算法 噪音 高斯分解 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]