Abstract:  A new algorithm was presented for the detection of corners of digital contour profiles based on barycenter. Firstly the length of curve on two sides of the each point on the contour profile is calculated to determine the support area, of which the end point is obtained with interpolation technique. Next an intersected polygons method is employed to calculate the barycenter of support area with given length. And then the distance between each point in the support area and its corresponding barycenter is used as feature strength to determine the feature point candidates. Finally the points with local maximum feature strength are chosen as corner point. The experiments show that the new algorithm is more accurate for corner detection and has stronger noisetolerant capacity than the other algorithms.
 

Key words: corner detection, support area, barycenter, planar curve, image processing