基于窄带M-S模型的猪序列切片图像的交互式分割  

Interactive Segmentation of Serial Pig Section Images Based on Narrow Band and Mumford-Shah Model

作　　者：杨长辉[1] 张绍祥[2] 王金勇 刘全利[1] 陈庄[1] 

机构地区：[1]重庆工学院计算机学院,重庆400050 [2]中国人民解放军第三军医大学,重庆400038 [3]重庆市畜牧科学研究院,重庆402460

出　　处：《计算机科学》2009年第1期234-235,238,共3页Computer Science

基　　金：重庆市重大科技攻关计划项目(编号:CSTC,2006AA1005)资助

摘　　要：提出了一种基于窄带M-S模型的图像交互式分割方法,解决了简化M-S的C-V方法针对图像全局,迭代速度慢和不能直接利用窄带法的问题;通过交互标记点和样条插值在目标附近建立初始曲线,以此曲线作为零水平集曲线,建立窄带,克服了窄带法局部求解的不足;实验表明,该方法大大提高了分割速度和精度,将此方法用于猪序列切片图像的分割,取得了较好的效果。An approach for interactive images segmentation based on narrow band scheme and Mumford-Shah model was presented, This solves the problem of C-V algorithm of Mumford-Shah model which is defined in the whole image region, results in low iterative speed and is unable to make use of narrow band directly. To overcome the shortages of local iteration method of narrow band, an original contour was established round the target based on interactive marker points and spline interpolation algorithm and narrow band of level set was constructed with the original contour. The experiment shows this method can greatly improve the segmentation speed and precision. This method is applied to serial pig section images segmentation,which gains preferable results.

关 键 词：图像分割 人机交互 窄带法 MUMFORD-SHAH模型 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]