An accurate and reliable circular coded target detection algorithm for vision measurement

Distributing coded targets on the measured object is a reliable and common method for achieving optimum target location and accurate matching of corresponding targets among multi-view images. The circular coded targets which based on a central circular target surrounded by a coded band is widely used in vision measurement. However, it is difficult to decode the coded target while the number of pixels in the coded band is small or the projection angle is large. Aiming at solve this problem, a detection algorithm using the gray gradient to get the central angles of each coded section was proposed. In this algorithm, an accurate ellipse detection which can get sub-pixel locations was adopted to extract ellipse centers, and some false ellipses whose error in the fit of best fit is large will be rejected. Then, gray gradients in the coded band are calculated to get the central angle of each coded section, and the coded target will be decoded accurately. The experiment results show that the algorithm can locate and identify coded targets accurately under complex measurement conditions.