Image object recognition is a process of matching between the image features extracted from an input image and the object models given. The image model is substantially the knowledge base about the object possessed by a machine, while image modeling is the process of description and is the foundation of recognition. Recognizing the 3-D object in an intensity remote sensing image by using a precise descriptive model is a task valuable for both theoretical research and practical usage, and is served to many fields of defense as well as civil economic construction. Considering the characteristics of this task, a 3D modeling approach of man-made objects based on generalized cones is presented in this paper. A man-made object can be modeled with one or more generalized cones. Except for basic parameters of a generalized cone, some new description parameters are defined for object searching and matching, which are generalized cone number, main element, length of the main element, elongation attribute, average gray and texture attribute of a surface, etc. In order to do exploitation and maintenance of an object recognition system and its object model base, model data is represented with the object-oriented method. It results in tree-like modeling structures of the data and is more suitable to remote sensing images. Based on the approach, a recognition system for specialized objects is developed and can be used for recognizing objects in real remote sensing images.%该文研究了用于遥感图像人造地物目标识别的三维建模方法.文中分析了识别任务的特点，比较了一般的建模方法，介绍了一种基于广义锥思想的几何表示方法，并利用面向对象的技术来表示模型内部数据及其操作.
展开▼
