A Corner Model and Optimal Filter

Patten recognition and photogrammetric tasks require models capable of detecting and measuring complex features normally found in all kind of natural and artificial scenes. Corners are the special features in images, and are of great use in computing camera calibration, object tracking and pattern reconstruction. Basically, a corner is defined as the junction points of two or more straight-line edges or curbs. In this paper, a novel optimal filter of interest point detection is presented based on a noise unit step edge function that is defined by the straight line edges. This filter is compared to other filters such as Gauss filter etc. The experimental results show that this optimal filter is robust.