采用非均匀的马尔科夫随机场模型来描述肖

像画与人脸图像之间的统计关系，并使用基于训练样本的非参数化的概率表示，在贝叶斯优化的框架下设计了迭

代采样算法，可以自动的从人脸图像生成特定风格的肖像画-W e learn the portraiture from a set of real artwork examples．Dif—

ferent from previous texture synthesis and image synthesis works that assumed modeling is homo—

geneous，Inhomogeneous M arkov Random Field Model is employed as the statistical mode