Cluster Matching of Artificial Targets in the Close Range Photogrammetry Based on the Epipolar Plane Constraint

We present a cluster matching method of artificial targets in close range photogrammetry based on the known points and epipolar plane constraint.Firstly,all the images are grouped according to the known points such as points on the AutoBar and codes.Secondly,image triplets per known point and their geometrical quality are calculated.Finally,parts of the image triplets with best geometric goodness are selected to match following the epipolar plane constraint.The experimental results show that more than 95% image points of about 10 000 in round 100 pieces of images are matched in less than 5 s,with matching error rates less than 0.1‰.