To solve the problem of low extraction precision of laser stripe centerline distributed in a non-Gaussian and non-uniform manner in stereo vision measurement,an extraction method of laser stripe centerline based on a Gaussian mixture model is proposed.First of all,the influence of laser speckle on the image is analyzed,and then laser speckle noise is effectively removed by mean filtering.Then position range of the laser stripes is located by threshold segmentation based on the OTSU method.Finally,the sub pixel centerline of laser stripes is extracted by Gaussian mixture distribution model.A Gaussian mixture distribution model can accurately describe the intensity distribution of laser stripes cross section,thus can achieve a high-precision extraction of extreme points at the center of the laser stripes.Experimental results show that this method can efficiently extract the centerline of the laser stripes.