A robust and accurate breakpoint detection method for line-structured laser scanner

Abstract

Line-structured laser scanner has been widely applied in industrial applications, such as 3D reconstruction. Breakpoint detection of the laser stripe centerline is necessary for further analysis. The laser stripe centerline is always not smooth in industrial environments due to the inevitable noises. In this paper, a robust and accurate breakpoint detection method named Fitting-Derivative-Optimization (FDO) is proposed, which mainly includes three steps. The first step is to fit the noisy laser stripe centerline. A novel piecewise fitting method without any empirical parameters is proposed in this step. The second step is to initialize the breakpoints. The second derivative curve of the point slope angle in the fitted laser stripe centerline is calculated. The local extremum points in the second derivative curve are extracted to be the initial breakpoints. The third step is to optimize the breakpoints. In this step, a new optimization method named Neighborhood-Circle (NC) is proposed, which optimizes each breakpoint in its neighborhood circularly to weaken the influence of correlation between breakpoints. Finally, robustness and accuracy of the proposed FDO method are evaluated by simulations and experiments respectively. In simulations, different Gaussian noises are added into the laser stripe centerline for comparisons. Experiments are also implemented to detect breakpoints for two industry parts by using a line-structured laser scanning system. Both of the simulation and experiment results demonstrate that the proposed FDO method can perform well in industrial environments.