摘要
二维Ostu方法同时考虑了图像的灰度信息和像素间的空间邻域信息,是一种有效的图像分割方法。针对二维Ostu方法计算量大的特点,采用含维向量变异量子粒子群算法,计算每一维的收敛度,以一定的概率对收敛度最小的维进行变异:让所有粒子在该维上的位置重新均匀分布在可行区域上,来搜索最优二维阈值向量,每个粒子代表一个可行的二维阈值向量,通过各个粒子的飞行来获得最优阈值。结果表明,所提出的方法不仅能得到理想的分割结果,而且计算量大大减少,达到了快速分割的目的,便于二维Ostu方法的实时应用。
2D Otsn method, which considers both the gray information and spatial neighbour information between pixels in image simultaneously ,is an efficient image segmentation method. However, its computational burden is very heavy. In light of this character, an optimization method, i. e. , a new Quantum-behaved particle swarm optimization with dimension vector mutation operator(QPSODMO) is utilised to calculate the convergent degree of every dimension. The dimension with minimal convergent degree is mutated according to a certain probability ;the positions of all particles in this dimension are distributed in the range[ - xmax,xmax] evenly for finding the best 2D threshold vector,in which each particle represents a possible 2D threshold vector and the best 2D threshold is obtained through the flying of every particle.Experimental results show that the proposed method can obtain ideal segmentation results and decreases the computation cost reasonably to achieve the goal of fast segmentation,and it is suitable for real time application of 2D Otsu method.
出处
《计算机应用与软件》
CSCD
2009年第5期228-231,共4页
Computer Applications and Software
关键词
图像分割
二维Ostu方法
粒子群算法
量子粒子群算法
Image segmentation 2D Otsu Particle swarm algorithm Quantum-behaved particle swarm optimisation