The use of FFT on sampled boundary distances
D.T. Wang, Y.N. Hew, Jeongwoo W. Lee, K.-Y. Chern, Chunpeng Wei, P.A. Ng · 2002
The fast Fourier transform (FFT) is applied to the sampled boundary distances (SBD) of an object image for recognition. The SBD is an ordered collection of samples of distances defined from an axis to points located on the boundary of the object image. With the FFTSBD approach, any object image that undergoes affine transformations such as translation, rotation, scaling, and stretching (along and/or perpendicular to the stretching axis) can be recognized. Therefore, the FFTSBD can be used to recognize an object at various locations, orientations, and distances from the camera and various objects of the same family that are stretched along an axis. FFTSBD can also be used to recognize heavily occluded objects, provided that the endpoints of the major axis are visible, and to recognize family of objects stretching along an axis, provided that the endpoints of that axis are visible. SBD can be used to extract axis of symmetry and to detect symmetry from a partially occluded object provided that the endpoints of major axis and axis of symmetry are not occluded.>