【摘要】 论文针对图像分割中存在的不确定性问题,通过研究不确定性人工智能中定性和定量的转换模型—云模型,提出一种新的基于云模型的图像分割方法。该方法采取交互式的方式选择训练样区,利用训练样区中的像素生成云模型,并通过泛概念树生成算法生成泛概念树,最后通过极大判定法判定像素所属类别,实现图像分割。这种方法能较好地描述图像目标的不确定性。通过几组实验,证明该方法可以准确地分割出目标,并优于传统的图像分割算法。更多还原

【Abstract】 This paper researches qualitative and quantitative transformation model-cloud model in uncertain artificial intelligence in order to solve uncertain questions which exist in image segmentation.We propose a new approach for image segmentation based on the cloud model.This method adopts the interactive way to choice training site,produce cloud models using pixels in the training site.Finally we produce a Pan-Concept-Tree with Pan-Concept-Tree Generation algorithm,and realize image segmentation with maximum likelihood principle.This method is more effective to express uncertainty in image.Several group of experiments show that this method can segment goals in image exactly,and more effectively than traditional image segmentation algorithm.更多还原