Centerline Extraction of Structured Light Stripe Under Complex Interference

A new structured light stripe centerline extraction algorithm is proposed. This algorithm can stably extract the stripe centerline under different interference environments and has the ability to repair broken lines. Using the unsupervised feature of the density clustering algorithm, even if there is a broken line or noise interference, it can still quickly cluster the set of candidate pixels on the centerline, greatly reducing the search range. Then, after experimental data analysis, it is found that the centerline of different scenes also has the features of high brightness and tend to be continuous. The energy function is defined by the Euclidean distance between pixels and the brightness value to model the feature, and the shortest path search algorithm is iteratively converged to accurately extract the centerline. The experimental results show that in three strongly disturbed structured light images, the algorithm extracts the centerline with a root mean square error of 0.1 pixel, the processing speed is 12.18 times faster than the image seam algorithm with the same strong anti-interference ability. The proposed algorithm greatly improves the anti-interference ability while maintaining fast operation speed and high accuracy, and effectively reduces the restrictions of using structured light measurement in industry.