Research on FPGA-based laser stripe center extraction algorithm

Laser scanning measurement systems are capable of accurately obtaining three-dimensional surface information of objects, with wide measurement range, high precision, and fast measurement speed, which are widely used in biomedical fields such as dental mold fabrication and surgical navigation. The laser triangulation systems are one of the most commonly used laser scanning systems. In laser triangulation systems, the processing speed of laser stripe images is crucial for determining the system's measurement efficiency. This paper focuses on real-time processing methods for laser stripe images in laser triangulation systems, developing a fast and accurate processing method based on the Zynq MPSoC platform. Gaussian filtering was first applied to the laser stripe images, followed by the calculation of eigenvalues and eigenvectors of the Hessian matrix. Subsequently, the rapid and accurate calculation of the laser stripe center position was completed. Experiments verified the accuracy and speed of the FPGA-based image processing algorithm proposed in this paper. Results showed that the accuracy of the FPGA-based algorithm is consistent with that of the traditional OpenCV-based algorithm, with measurement time reduced to 1/18 of that on the software platform. This significantly improves the speed of laser stripe center extraction without compromising image processing accuracy. The FPGA-based laser stripe center extraction algorithm developed in this study is valuable for enhancing the measurement speed of laser triangulation systems and offers significant insights for high-precision, fast image processing technologies in related fields.