基于二类切比雪夫正交多项式非参数混合模型的图像分割  

Image Segmentation Based on Non-Parametric Mixture Models of Chebyshev Orthogonal Polynomials of the Second Kind

作　　者：刘哲[1,2] 宋余庆[1] 陈健美[1] 谢从华[1] 宋旼珊[3] 

机构地区：[1]江苏大学计算机科学与通信工程学院,江苏镇江212013 [2]吉林师范大学计算机学院,吉林四平136000 [3]江苏科技大学数理学院,江苏镇江212013

出　　处：《计算机研究与发展》2011年第11期2008-2014,共7页Journal of Computer Research and Development

基　　金：国家自然科学基金项目(60841003);江苏省软件与集成电路专项基金项目(2009[100]);江苏省博士创新基金项目(CX10B_274Z)

摘　　要：有参混合模型需要假设模型为某种已知的参数模型,而实际数据往往很难假设出这种参数模型的分布.为此,提出一种二类切比雪夫正交多项式的非参数图像混合模型分割方法.首先,设计出一种基于二类切比雪夫正交多项式的图像非参数混合模型,每一个模型的平滑参数根据误差方法和最小的准则进行计算.然后,利用随机期望最大(SEM)算法求解正交多项式系数和每一个模型的权重.此方法不需要对模型作任何假设,可以有效克服有参混合模型与实际数据分布不一致的问题.实验表明,该方法比高斯混合模型分割效率更高,并比其他非参数正交多项式混合模型有更好的分割效果.To solve the problem of over-reliance on priori assumptions of the parameter methods for finite mixture models, a nonparametric mixture model of Chebyshev orthogonal polynomials of the second kind for image segmentation method is proposed in this paper. Firstly, an image nonparametric misture model based on Chebyshev orthogonal polynomials of the second kind is designed. The mixture identification step based on the maximisation of the likelihood can be realised without hypothesis on the distribution of the conditional probability density function（PDF）. In this paper, we intend to give some simulation results for the determination of the smoothing parameter, and use mean integrated squared error （MISE） estimation of the smoothing parameter for each model. Secondly, the stochastic expectation maximum （SEM） algorithm is used to estimate the Chebyshev orthogonal polynomial coefficients and the model of the weight. This method does not require any priori assumptions on the model, and it can effectively overcome the ＂model mismatch＂ problem. The algorithm finds the most likely number of classes and their associated model parameters and generates a segmentation of the image by classifying the pixels into these classes. Compared with the segmentation methods of other orthogonal polynomials, this new method is much more fast in speed and better segmentation quality. The experimental results about the image segmentation show that this method is better than the Gaussian mixture model segmentation results.

关 键 词：非参数混合模型 图像分割 平滑参数 正交多项式 概率密度函数 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]