For the precise calibration of laser triangular displacement detection based on position sensitive devices (PSD) with the influence of complex photoelectric noise, a joint calibration method based on multivariate adaptive Kalman filter (MAKF) and non-uniform B-spline is proposed. The nonlinear characteristic of displacement sensor is shown in detail by analyzing the measurement principle and computing methods. Aiming at the nonlinear problem, the method of curve fitting for the calibration is given. In consideration of the characteristic of curve fitting, in this paper, a preprocess method called MAKF is proposed for the photoelectric noise elimination. Then, the B-spline curve fitting model is built by dividing the knot vector nonuniformly, which makes the precision of the sensor further promoted. Calibration experiments are carried out under actual working conditions. The experimental results show that the curve fitting calibration method can complete the high-precision calibration of the laser displacement sensor. The positioning accuracy is 0.7%, the average measurement error can reach 0.012 mm, and the calibration mean square error is 2.12×10 -5 mm 2.