Abstract: System parameter calibration is crucial to structure light three-dimension measurement system. The calibration error is mainly due to the detection accuracy of the characteristic circular center of calibration board and the gamma distortion of the projector and camera. An accurate circular center detection method and a gamma pre-calibration method were proposed to improve calibration accuracy. The pixel edge detection with Sobel operator and subpixel edge detection with orthogonal Fourier-Mellin moments (OFMM) operator were used to detect the circular edge accurately, moreover, the circular center was detected by ellipse fitting method. The gamma nonlinear distortion model was analyzed to obtain the system gamma value, whose reciprocal was used to modify the ideal sinusoidal fringe pattern as its exponent. The experimental results show the effectiveness of the methods. The calibration accuracy increases by a factor of 3.5 in X direction and 5 in Y direction compared with that without these methods.