Abstract
A technique for extracting multiline lasers is proposed for handheld scanner. First, a robust matching method for multi-lasers is presented in this work to distinguish every laser stripe in the left and right images by using back-projection algorithm. Then, an improved DBSCAN (Density-Based Spatial Clustering of Applications with Noise) clustering method is promoted to detect noise in the laser images caused by complex scenarios. This approach is focused on the features of the laser stripes in the images, such as normal vectors, stripe width and gray values. These parameters are used as the criteria to find the noise in the images. After the calibration results of stereo-vision and laser-matching results are combined, the 3D information of the noise parts in the laser stripes are recovered accurately. Finally, a subpixel interpolation method based on Shepard interpolation is used to achieve the precise subpixel centerline extraction of laser stripes. And the experiments further valide the effectiveness of the proposed method.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call