A new approach to generate conic section approximation for object boundaries is presented. The algorithm takes advantage of the geometric features of conic sections, such as the chord, the characteristic point, and the guiding triangle properties, to assist in formulating the conic sections. A generalized guiding triangle model is applied to extract the dominant features and then to derive the parameters of conic sections from a sequence of edge pixels. The approach has an O(nlog n) computational complexity which is lower than that most other conic fitting procedures.
ASJC Scopus subject areas
- Signal Processing
- Computer Vision and Pattern Recognition