Abstract: In the process of local imaging detection, due to the difference of parts with complex outline or placement state, the parts and image plane coordinates have a certain angle, which can cause the problem that there are asymmetric point sets or no symmetric points in the acquired symmetric sets. When using traditional Hough transform or fitting method to detect assembly coaxality, it has much error. A detection algorithm of assembly coaxality in local imaging was proposed to solve the above problem.This algorithm firstly extracts the upper and lower edge point sets and then counts the parameter space points projecting in Hough space from two point sets based on Hough linear transform. Finally, it searches the maximum point of parameter points cumulative number, the point corresponds to the symmetric axis which is the optimal symmetric axis. The simulated result shows that the method can extract the optimal symmetric axis with high accuracy, and the coaxality error is only 0.002 7 . Therefore, the algorithm of assembly coaxality in local imaging is feasible.