A Two-Dimensional Sub-Pixel Edge Detection Algorithm Based on Hyperbolic Tangent

Locating edges in images accurately is important to improve the precision of visual measurement. Firstly, from the characteristics of edge gray distribution in actual images, a method is put forward to detect sub-pixel edges, which fits edge gray distribution with hyperbolic tangent function, and gets subpixel edges through least-square method. Then, the accuracy of the algorithm is evaluated, taking measurement error of sizes of high-accuracy checkerboard as criterion. This evaluation combines with camera calibration technique, which considers factors of lens distortion and perspective distortion.