A research on improved canny edge detection algorithm

Abstract

Classical Canny Operator plays an important role in the image edge detection. The paper analyses the theory of the traditional Canny edge algorithm and does some improvements on the parts of smoothing filter selection, point amplitude calculation, and high or low threshold selection. The improved Canny algorithm uses B-spline function instead of Gaussian function; calculates gradient amplitude in 3x3 neighborhoods; and selects thresholds on the basis of gradient histogram. The experiment proves that the new algorithm improves the accuracy of positioning and provides a better and evident de-noising effect. © 2011 Springer-Verlag.