Abstract: Distributing reference points on the object to be measured is a reliable and common method for achieving optimum target location and accurate correspondence among multi-view images. In this paper circular uncoded targets and coded targets are used as reference points, and an algorithm of automatic reference point detection is proposed. Targets are extracted from the images according to their size, shape, intensity, etc. An improved method to identify every coded target is proposed. The gray scale centroid algorithm is applied to get subpixel locations of both uncoded and coded targets. Practical examples show that the algorithm can identify and locate reference points accurately. It is robust to the change of projection angles and noise.

Key words: close range photogrammetry, ellipse detection, sub pixel, coded target, gray scale centroid