摘要
几何不变量,特别是射影不变量,是基于单视点灰度图像识别三维物体的一条有效途径.但理论研究表明,只有特定的几何约束结构,才具有射影不变量.所以,研究并发现这种几何约束结构就具有十分重要的意义.该文提出了一种新的由相邻3平面上5条直线组成的几何约束结构及其所具有的射影不变量.该结构较Sugimoto提出的几何约束结构简单,可从结构同样复杂的物体中获得更多的几何不变量,有利于提高物体识别的稳定性;同时,由于该结构大量存在于由多面体组合而构成的人造物体及地面建筑物中,因此它非常适合这类物体的识别.实验验证了文中提出的几何约束结构具有不随物体成像视点改变的射影不变量.
Geometric invariant is an effective tool in recognizing 3D object from a single gray image. However, researches show that only some special geometric constrained structures possess projective invariants. Therefore, it is necessary to study and find out more such structures. A new geometric constrained structure with three planes and five lines and its projective invariant are proposed. This structure is simpler than the Sugimoto＇s structure of six lines on three planes. So, more projective invariants can be extracted from the same object, which can improve the stableness of recognition. At the same time, because the new structure widely exists in the artificial polyhedral objects and buildings, it is more suitable for the recognition of such kinds of objects. The experimental results show that the proposed structure has projective invariant which does not change with the variation of the viewpoints.
出处
《计算机学报》
EI
CSCD
北大核心
2005年第10期1740-1744,共5页
Chinese Journal of Computers
关键词
三维物体识别
几何约束结构
射影变换
射影不变量
单视点灰度图像
3D object recognition
geometric constrained structure
projective transformation
projective invariant
single-view gray image
