文章从多传感器图像的属性研究出发,提出图像分割的新概念。建立图像邻域系统的Markov过程和基于Markov过程的Dempster-Shafer融合分割模型。并充分利用象素间的空间邻接关系。用证据理论描述图象分类的不确定性,较好地解决了传感器图像分割信息不全的问题,实现了景物图像的准确分割。
This paper proposes the new concept of image segmentation by studying the properties of multi-sensor images We define a general Dempster-Shafer evidential Markovian field model-based segmentation algorithms that can take into account the spatial interactions between pixels, and extend the classical Bayesian Hide Markovian Model to model the resulting uncertainty with the theory of evidence, which can improve the accuracy of image segmentation.