基于最大熵和粒子群优化的红外图像分割

摘要

Maximum fuzzy entropy has been proven to be an efficient method for image segmentation. The key problem associated with this method is to find the optimal parameter combination of membership function so that an image can be transformed into fuzzy domain with maximum fuzzy entropy. However, it is computationally intensive or even impossible to find the optimal parameter combination of membership function directly.A new optimization algorithm, namely, PSO algorithm is used to search optimal parameter combination. Each particle represents a possible parameter combination. A swarm of particles are initialized and fly through solution space for targeting the optimal solution. The proposed method is used to segment infrared image and ideal segmentation results can be obtained with less computation cost.%最大模糊熵是一种有效的图像分割方法,该方法的一个关键问题是确定模糊隶属度函数的最优参数组合,从而使得图像变换到模糊域后的模糊熵最大.但是直接采用穷举法来寻找最优参数组合的计算量是很大的,甚至是不可能的.因此,提出了采用一种新的优化方法,即粒子群算法来寻找最优参数组合.在参数搜索空间中,随机初始化一群粒子,通过粒子之间的相互协作来寻找最优解.所提出的方法用于分割红外图像的结果表明,花费很小的计算代价就可以获得理想的分割结果.