In this paper, we propose a source optimization method of lithography tools based on quadratic programming. The fidelity of the aerial image to the mask object pattern is designed as the object function. Then by using the linear relationship between the aerial image and point sources, the source optimization is transformed to a quadratic programming problem. The mask object pattern is divided into a limitation area and a comparison area. The constraint condition is applied in the limitation area and the object function is applied in the comparison area. The one-dimensional (1D) isolated space and two-dimensional (2D) contact hole array patterns are used to validate our method. The impacts of the threshold and focus on the optimization results are studied. According to the simulation results, the global optimization source is obtained by using the proposed method. The image quality is improved and the process window is increased after the source optimization.