In view of the problems of slow manual detection speed and poor precision in the detection of the dimensions of irregular workpieces, an automatic measurement method of inner concave polygon based on point cloud data density extraction is proposed. First of all, the two-eye structure light scanning method is used to obtain the original data of the point cloud on the surface of the part. Secondly, using KD-tree to establish point cloud topology relationships and to perform radius filtering. Thirdly, using greedy algorithm to rebuilt mesh surface. Finally, the base plane bias and density extraction are used to realize the edge feature point and noise classification of the point cloud, and the dimension characteristics of the inner concave polygon aperture are extracted and the size is calculated. The algorithm is verified by the curved automobile dashboard with the characteristics of the inner concave hexagon. The experimental results show that the algorithm can extract features with limited degree and the accuracy of internal feature size measurement is 60 μm. Compared to traditional measurement methods, a single inner concave hexagonal measurement time can be reduced to 1 s, enabling automated and fast dimensional measurement of irregular workpieces.