Linear boundary detection and corner detection are major challenges in computer vision. There exist many solutions for these problems based either on edge detection or interpolation methods but they are inexact in the sense that they do not talk of bounds. The basic objective of the study in this paper is to find out how exactly we can locate or restore linear boundaries or corners in the real plane, rather than pixel domain, by observation at discrete pixels. This paper devises new algorithms for linear boundary detection and corner detection using computational and digital geometric techniques.
展开▼