一种基于高斯混合模型的改进EM算法研究

Advanced EM algorithm based on Gaussian mixture model

摘要: 针对传统EM算法存在估计参数不具有最优性，以及在参数估计中需要人工参与等问题，提出一种基于高斯混合模型的改进EM算法。采用无人工参与的无监督思想，获取高斯混合模型对直方图拟合的最优参数组合。实验表明，该算法不仅能够快速地估计模型参量，而且能够给出最优参数，并在图像增强中使细节更明显，对比度更适中。

Abstract: In order to solve the disadvantages of traditional expectation maximization (EM) algorithm which lacks parameters optimization and needs human operation when estimating parameters, an improved EM algorithm based on Gaussian mixture model was proposed. The unsupervised theory was used to calculate optimal Gaussian mixture model parameters. The subjective and objective indices of experiments show that the algorithm can not only estimate parameters quickly but also figure out the optimal parameters, making the detail more obvious and the contrast more moderate in image enhancement application.