摘要
为了更好地分割图像,对传统Split-merge算法作出改进:PCNN先进行分裂,用一种简化的Mumford-Shah模型进行合并,使得分裂阶段不仅无效分割减少,而且无方块效应,对边缘定位准确;合并阶段能够理想地将分裂后的区域合并为感兴趣的前景与不感兴趣的背景,误合并与欠合并大幅减小。对Papav,Monkey,Twoman图像的仿真表明,其分割结果和运行时间均优于AMS,MBMS算法。仿真结果证明,本文算法是一种适用于图像分割的有效算法。
In order to segment images more effectively,an improved Split-merge algorithm is proposed.Images are splited by PCNN and merged via a simplified Mumford-Shah model.Its advantages are as follows: at the split stage,it can reduce the number of invalid split and is immune from the block-effect,and locate more accurately edges;at the merge stage,it can merge areas into interesting foregrounds and uninteresting backgrounds and reduce the chance of the incorrect and insufficient merge.Experiments demonstrate that the algorithm is superior to both AMS and MBMS algorithms on the segmentation effect and the computation time.Thus,the method is effective for the image segmentation.
出处
《数据采集与处理》
CSCD
北大核心
2010年第3期353-357,共5页
Journal of Data Acquisition and Processing
基金
国家自然科学基金(60872075)资助项目
高等学校科技创新工程重大项目培育资金(706028)资助项目
江苏省自然科学基金(BK2007103)资助项目
东南大学优秀博士学位论文基金(YBJJ0908)资助项目
关键词
图像分割
分裂合并法
脉冲耦合神经网
image segmentation
Split-merge method
pulse coupled neural network