A control point extraction and optimization method for laser scanning projection graphics based on feature adaptation was proposed. It achieved the adaptive distribution of projection control points for different graphics features by dividing the feature region and non-feature region of the to-be-projected graphics according to the curvature of each adjacent point and the normal vector angle. Meanwhile, to ensure the shape accuracy of the projected graphics after projection control points were diluted, we also studied the dilution and optimization of the number of control points based on the spatial resolution of human eyes. Finally, the optimized control points of the projected graphics were used as the scanning projection points to complete the cyclic scanning projection of the to-be-projected graphics contour. The experimental results show that the proposed method, when applied to a laser scanning projection system, can reduce the number of projection control points by more than 20% and solve the “flicker” problem in the scanning projection of complex graphics.