基于微粒子群理论的二维最大类间方差阈值分割算法

The Two-dimensional Maximum Between-cluster Variance Threshold Segmentation Algorithm Basing on Particle Swarm Optimization

针对传统二维最大类间方差(Otsu)阈值分割算法处理红外图像耗时的缺点,提出了一种应用微粒子群理论的二维Otsu阈值分割算法。该算法利用粒子群理论的群体智能的特点,通过优化得出粒子的个体极值和全局极值,并根据这两种极值来更新粒子的位置和速度以获得最佳的分割阈值向量。通过对算法中惯性权重和学习因子的讨论确定了最佳的参数选择方案。仿真结果表明,该算法计算准确,流程简单,其运行时间仅为原始算法的5%左右,是一种快速有效的图像阈值分割算法。

The traditional two-dimension maximum between-cluster variance algorithm is always time-consuming in processing infrared image. A kind of using particle swarm optimization in two-dimension maximum between-cluster variance algorithm is proposed. This algorithm, using the character of community intelligence, obtains the individual extremum and overall extremum, then, according to the two extremums to update the particle＇s position and velocity to get the best division threshold vector. The discussion of inertia weight and study factor has determined the best parameter selection scheme. The simulation result indicates that this algorithm is correct, the flow is simple, and its running time is only 5% of that of the original method. It＇s a kind of fast effective image threshold segmentation algorithm.