Abstract: A model of aperture measurement based on quadratic curve invariants was proposed to deal with the low detection efficiency and the problem that detection accuracy was greatly affected by the environment. The conic invariant aperture of the hole was obtained by ellipse geometry fitting on the surface of the hole. Firstly, the world coordinate system was established and the transformation relationship between the world coordinate system and the camera coordinate system was determined by calibration. The improved candy algorithm was used to extract the sub-pixel coordinates of the hole edge, and the sub-pixel coordinates of the hole edge were projected onto the hole surface according to the coordinate transformation relationship. Then, the ellipse geometric fitting of the hole edge curve on the hole surface was performed to obtain the quadratic curve invariant. Finally, the conic invariant was used to measure the aperture on the test bench, and the results were compared with those of the inner micrometer to verify the accuracy of the conic invariant model. 

Key words: aperture measurement, coordinate transformation, geometric fitting, conic invariant