Abstract: Aiming at the problem that the extraction accuracy of laser stripe center can be easily disturbed by a large amount of noise in line structured light 3D shape measurement, a method for extracting the stripe subpixel center was proposed. First, the noise in the stripe image was analyzed and preprocessed by the average method and median filtering. Then, the stripe was obtained by iterative threshold segmentation and morphological method, and the pixel center was extracted by distance transform. Finally, on the basis of the pixel center, binary information and light intensity, the precise subpixel center of the stripe was extracted by combining curve fitting with barycenter method. After simulation analysis and experimental verification, the maximum value and average value of column coordinate deviation between adjacent rows of stripe center are about 2 pixels and 0.3 pixels, respectively, smaller than the traditional method. Experimental results show that, the algorithm based on the effective use of the stripe gray distribution, can effectively reduce the influence of noise, approach to the true center of stripe more cosely, with extremely strong noise proof ability.