In order to improve the accuracy of flatness measurement, an optimized point laser method is proposed. Firstly, point laser measurement coordinate system is converted. Secondly, plan is calibrated with laser beam, and mixed noise points of laser acquisition are reduced using sliding filter to avoid affecting the test. Nonlinear error of measurement data and output model of laser displacement sensor are established to compensate the output error. Finally, the best fitting plane is obtained by least square algorithm. Experiment results show that the average flatness error of the optimized point laser is 5.189 μm, which is 18.77%, 16.78%, 12.45%, 9.17% and 6.54% lower than direct method, plane scanning, transverse scanning, point laser and spectral confocal, and the measurement accuracy is higher than other algorithms.