图像的二维自回归滑动平均模型

2D Mixed Autoregressive Moving Average ModelofImage

在一维自回归求和滑动平均模型（ＡＲＩＭＡ（ｐ；ｄ；ｑ）的基础上，探讨了其在二维图像空域序列的模型推广．讨论了图像的二维自回归滑动平均模型ＡＲＭＡ（ｍ ，ｎ；ｐ，ｑ）的建模方法，包括利用模型的自相关和偏相关函数的截尾性及Ｆ检验准则对模型定阶，采用两步估计法进行自回归模型（ＡＲ（ｐ））和滑动平均模型（ＭＡ（ｑ））的参数估计．将此模型应用于对一类纹理图像的建模中，在较低阶时就能得到较好的定量描述，从而为图像的纹理特征提取提供了分类依据．

On the basisof1 Dautoregressive integrated m oving average m odelARIMA (p;d;q), the 2 D m ixed autoregressive m oving average m odel ARMA (m , n; p, q) w as developed. The m ethod for establishing m odel, including the order determ ination (according to the cutoff property of the autocorrelation coefficientand the partialautocorrelation coefficientofthe m odeland the Fprinciple) and the param eter estim ation (by appling 2 step estim ation w ith the coefficientofARand MA m odel), w as presented. Finally, this m odelwas successfully applied in the texture im age m odeling, and itachieved a good quantitative description with such im age. Itis benefitfor the analysis oftexture im age by using this m odel.