Line-Structured Light Based 3D Reconstruction Algorithm for Ice Surface

Line structured light has a wide range of applications in 3D measurement, such as metal parts, weld seams and other precision shaft parts, and the surface of ice cubes has similar characteristics and can also be measured in 3D using line structured light. However, due to the distinctive optical characteristics of the ice surface, the brightness of the captured laser stripes at certain positions may be either insufficient or excessively intense, posing difficulties in accurately identifying the centerline of the laser stripes; and the high transparency of the ice surface, part of the light will be projected into the interior of the ice resulting in a weakened reflection, and the light projected into the interior of the light will also diffuse and affect the imaging. Therefore, how to accurately measure the three-dimensional structure of the surface has become a topic to be studied. To accurately identify the centerline of the laser stripes within the interference image, this paper introduces a laser stripe center extraction algorithm that combines deep learning with the Steger algorithm. By leveraging the robust feature learning capabilities of deep neural networks, the algorithm can effectively capture the overall characteristics of the laser stripes on the ice surface, enabling precise detection and segmentation of the stripes. Subsequently, the Steger algorithm is applied to determine the centerline of the laser stripes with high accuracy. Subsequently, Steger's algorithm is employed to precisely determine the centerline of the laser stripes. This approach effectively mitigates the problem of not being able to accurately extract the center of the laser stripes caused by the high transparency of ice and complex surface reflection characteristics. Subsequent experiments have proved that this method can greatly reduce the extraction time of laser stripe center and significantly improve the accuracy of laser stripe center extraction and 3D reconstruction.