粒子群优化增强大津法的气泡分割方法研究  

Study on Bubble Segmentation Method Based on Particle Swarm Optimization & Enhanced Method of OTSU

作　　者：周红娟[1] 周云龙[2] 

机构地区：[1]东北电力大学自动化工程学院,吉林吉林132012 [2]东北电力大学能源与动力工程学院,吉林吉林132012

出　　处：《东北电力大学学报》2011年第1期69-74,共6页Journal of Northeast Electric Power University

基　　金：国家自然科学基金(50976018)

摘　　要：提出了粒子群算法优化增强大津法来实现气泡图像的快速准确的分割。首先介绍了阈值分割中的直方图法、经典大津法、迭代法的阈值选取原理,并利用这三种方法对垂直上升管气液两相流中稀疏上升气泡图像进行了分割,通过效果比较,并结合气泡图像的特点,提出了一种以粒子群算法优化增强大津法的图像分割法,然后利用粒子群算法的全局搜索能力改善增强大津法的阈值选取时间,求出分割阈值完成气泡的分割。实验结果表明,此方法能更准确,更快速的实现稀疏气泡的分割。Bubble image segmentation was achieved accurately and quickly by using particle swarm optimization algorithm and enhanced OTSU method.Firstly the principles of petronas method,conventional method of OTSU and iteration method were introduced,secondly using three of them to segment images of rising sparse bubbles of Gas-liquid two phase flow in vertical tube,finally the three kinds of image segmentation method were compared,and a image segmentation method was proposed by using particle swarm algorithm to optimize the enhanced method of the OTSU based on the characteristics of bubble image,and then using the global search ability of particle swarm optimization algorithm to improve the direshold value time of the enhanced OTSU method,get thresholds and complete bubble segmentation.The experimental results showed that the new method can obtain more accurate and quickly impact of sparse bubble image segmentation.

关 键 词：图像分割 稀疏气泡 粒子群算法 增强大津法 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]