Abstract: Light-stripe matching is one of the difficulties of raster binocular vision sensor. Usually, matching methods based light-stripe coding require projecting multi-images or viewing minimum stripes in one image. Additionally, when the depth of the measured object has a relatively large range of variety, mismatching will frequently occur. To solve this problem, a new method was presented as follows: The raster binocular stereo vision sensor was regarded as two raster structured-light vision sensors which had been calibrated under the same measurement coordinate system. The 3D coordinates of the light-stripe center points was measured by the two rater structured-light vision sensors respectively. The distances between the light-stripe center point sets were calculated and the ones which had the minimum distance and the same model number were the matching light-stripes. The experiment results indicate the validity of the presented method for light-stripes matching.