Improved Steger Line Laser Stripe Centerline Extraction Algorithm

The extraction of the central line of laser stripes is a crucial technique in three-dimensional measurement. To overcome the limitations of existing algorithms for central line extraction, particularly in terms of speed and accuracy, this paper proposes an enhanced version of the Steger algorithm that incorporates parallel-accelerated Gaussian filtering and adaptive Gaussian templates. Initially, an adaptive Gaussian template is utilized, wherein the width of the light stripe at various positions is estimated to generate the Gaussian template. Subsequently, Gaussian filtering is applied to improve the accuracy of central line extraction. This method effectively mitigates the impact of surface contamination and reflection artifacts. Furthermore, the algorithm enhances processing speed through parallel execution. Experimental results indicate that the enhanced algorithm extracts central lines that are smoother, more coherent, and more accurate. By computing the standard deviation of central line extraction across various image types, it is observed that the error in the enhanced Steger algorithm has decreased by $16.6 \%$ on average.