Vision research at TI: 1994-95 progress

被引:0
|
作者
Rao, KS
Flinchbaugh, B
机构
关键词
D O I
暂无
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
This paper describes recent vision research progress at Texas Instruments (TI). Our research for surveillance has primarily focussed on curved 3-D object description from single aerial images. For this we use shadow information to infer the shape of the objects describable as straight homogeneous generalized cones (SHGCs). This year we concentrated on inferring the cross-section of the generalized cones. Here we highlight the finding of a constant of 54.73561... similar or equal to 54.74 degrees in the projection geometry of circles. We obtained expressions for the sweep rule of SHGCs like the solid of revolution, and we developed and tested a system called SHADOW for description of curved 3-D objects such as oil tanks from aerial images. This paper also describes vision research progress at TI in object-based video compression and automatic video indexing, with potential applications for security monitoring and battlefield awareness.
引用
收藏
页码:173 / 175
页数:3
相关论文
共 50 条