Projective Invariant Recognition

Object recognition is one of the important parts of computer vision, and has been widely applied in the fields such as military affairs and automatization. Description and recognition of plane has been processed depend on the geometry model of continental group and affine group anciently, with the development of the computer vision application, people attaches importance to the investigation of the perspective invariants, thus the investigation of perspective invariants gradually become an important problem in the computer vision.A new kind of method for recognition of planar object is presented based on steerable filters. The proposed method consists of two parts: Firstly, with the help of steearable filters, we detect the orientations of the planar object edges, extract some geometry characteristics invariant under the projective transformation, construct the canonical frame. And then, we recognize planar objects by moments of the filled canonical frames. This method is local method, and can be appliced to objects with some partial occlusion.