Abstract：This paper presents a method that provides an applicable device to obtain 3-Dinformation for a scene and to make 3-D description of the scene objects.It makes use of fuzzy entropy in a whole image as (?)lustering evaluation functionto obtain better region segmentation.We use moments and functions as the characteristies of the regions and form asymbol set to represent the contents of the image.The 3-D coordinates of each ele-ment(region)will be calculated through geometric relationships between stereoimage couple.According to these 3-D information of the elements in different space and time,one can get their 3-D moving statements by connecting 3-D statements of the ele-ments,we bring about 3-D detection and description.
 Bezdek J C, Pattern recognition with fuzzy objective function algorithm, Plenum Press, 1981
 董再励,彭克敏.基于分类熵的图象分割.1986年全 国模式识别和人工智能会议论文集
 Jain R C, Segmentation of frame sequences obtained by a moving observer, IEEE Trans PAMI, l984;(5):624-629