CAJ | 학술논문

为了提高区域生长的分割精度和模糊连通度算法的运算速度,减少算法所需的人工干预和种子点选取对分割结果的影响,提出了一种融合区域生长和改进模糊连通度,并结合置信区间和区域竞争方法,用于医学图像分割的动态相对模糊区域生长算法.首先算法使用置信区间区域生长快速地得出初步的多个对象分割结果,然后利用置信区间的参数,对分割区域间重叠部分使用动态相对模糊连通度算法进行再分割.通过在大量医学图像上的实验,实现了复杂背景下的图像分割.实验结果表明,该算法所需交互少,并能提高分割精度和速度.
위료제고구역생장적분할정도화모호련통도산법적운산속도,감소산법소수적인공간예화충자점선취대분할결과적영향,제출료일충융합구역생장화개진모호련통도,병결합치신구간화구역경쟁방법,용우의학도상분할적동태상대모호구역생장산법.수선산법사용치신구간구역생장쾌속지득출초보적다개대상분할결과,연후이용치신구간적삼수,대분할구역간중첩부분사용동태상대모호련통도산법진행재분할.통과재대량의학도상상적실험,실현료복잡배경하적도상분할.실험결과표명,해산법소수교호소,병능제고분할정도화속도.
To improve the segmentation accuracy of region growing and the speed of fuzzy connected;reduce the number of user interaction and the influence of placing a seed point,a segmentation algorithm of combining region growing with improved fuzzy connected based on confidence interval and region competition is presented.The approximate segmentation result of multiple objects is obtained by region growing of confidence interval rapidly.And the parameters of confidence interval are used for dynamic relative fuzzy connected to segment the intersecting region of the foregoing result.The segmentation from image of complex background is achieved by plentiful experiment of medical images.The experimental results illustrate that the proposed algorithm improve the accuracy and the speed of segmentation with very little interaction.