A Laser Centerline Extraction Method for 3D Reconstruction of Throttle Valves

In the process of 3D measurement and reconstruction using line-structured light, the most critical step is the extraction of the laser stripe centerline, as its extraction accuracy has a decisive impact on subsequent measurement precision. Due to its metallic reflectivity, the throttle valve can cause the laser stripe to lose its original Gaussian characteristics, thereby reducing extraction accuracy. This paper proposes a method for extracting the stripe centerline specifically for the 3D measurement of throttle valves. First, the method preprocesses the grayscale image using filtering to smooth the image and reduce interference from noise and speckles. Next, the image is binarized to reduce its complexity. Then, edge detection is used to extract the stripe contour. Subsequently, the light stripe contours are extracted through edge detection, and the centerline of the laser stripe is extracted using the gray level centroid method for the detected laser stripe contours. The comparison results yield that the proposed method improves the speed and accuracy of stripe center extraction compared to traditional grayscale centroid methods, meeting the requirements for real-time 3D reconstruction and ensuring the precision and real-time performance of 3D measurement and reconstruction.