基于高斯混合模型的视频对象分割算法

针对应用高斯混合模型（GMM）进行视频建模与分割时的模型选择及参数估计初值选择的难点，提出了一种基于GMM的视频对象分割算法．首先进行特征提取，在特征矢量中引入加权运动信息，可根据不同需要选择合理的加权系数，然后通过分割投影进行模型选择及期望最大化（EM）算法的参数初始化并估计参数，这种初值选择方案使得EM算法的初值和真实值较接近，加快了迭代运算的收敛速度，从而提高了视频对象的分割速度，最后对特征矢量进行聚类分割．仿真实验表明，在保持良好分割效果的同时，所提算法的运算速度约为常规方案的76％，并且具有良好的稳定性．

Video Object Segmentation Based on Gaussian Mixture Model

Focusing on the problems of model selection of Gaussian mixture model(GMM) and parameter initialization of the expectation maximization(EM)algorithm,a novel video object segmentation algorithm based on GMM was proposed.The number of mixture components of GMM is estimated and the EM algorithm is initialized through segmentation projection after extracting feature.Then the EM algorithm is applied to estimate the distribution of feature vectors.Finally the segmentation is carried out by clustering each pixel into appropriate component according to maximum likelihood criterion.According to different applications,this algorithm can choose properly weighted coefficient by introducing the weighted motion information to feature vectors.The proposed algorithm can greatly accelerate the convergence of the EM algorithm since the initial value approximates its real value.As a result,the speed of the video object segmentation is improved.Experimental results demonstrate that the proposed method can extract moving objects from video sequences successfully.The algorithm proposed is more stable and the computational complexity is about 76% of the original scheme.