The absorption and refraction of light by water made high-precision 3D (three-dimensional) reconstruction of underwater concrete a challenging task. This paper proposed a 3D reconstruction method combining line structured light and stereo vision. To improve the reconstruction accuracy, the epipolar constraint was introduced in the light plane calibration process to limit the fringe noise data during calibration matching. A color camera and a monochrome camera were used simultaneously to characterize the real underwater 3D environment. After matching the left and right images, the color information of the color image was retained, and the color information of the point cloud was enhanced. Finally, experiments were conducted in a water tank, and the results indicated that the 3D reconstruction error for underwater concrete was 4.48 %. Moreover, the color enhancement of the point cloud achieved the highest overall scores across the four no-reference underwater image quality assessment metrics.

中文翻译：

---

使用集成线结构光和立体视觉对水下混凝土进行高精度 3D 重建


水对光的吸收和折射使水下混凝土的高精度 3D（三维）重建成为一项具有挑战性的任务。本文提出了一种结合线结构光和立体视觉的 3D 重建方法。为了提高重建精度，在光平面标定过程中引入了 epipolar 约束，以限制标定匹配过程中的条纹噪声数据。同时使用彩色相机和单色相机来表征真实的水下 3D 环境。左右图像匹配后，保留了彩色图像的颜色信息，增强了点云的颜色信息。最后，在水箱中进行实验，结果表明水下混凝土的三维重建误差为 4.48 %。此外，点云的色彩增强在四个无参考水下图像质量评估指标中获得了最高的总分。