针对水下三维重建成像时图像失真的问题, 提出一种基于非平行双目视觉系统的转化条件和转化模型, 将摄像机透过平面防水罩拍摄的水下图像转化为摄像机与目标间无水存在的一般空气图像.当摄像机焦点在水与空气分界面处时, 水下图像和空气图像像素间存在一一对应关系, 当摄像机焦点位于水面外时, 该模型在物距远大于焦距及焦点沿z轴方向与折射面距离的情况下仍成立.实验利用特征点三维坐标重建验证模型的正确性, 结果显示, 利用本文方法重建的三维坐标和实际三维坐标相比, x、y、z轴方向的重建误差平均值分别为2.23%, 1.51%, 1.10%, 表明该模型转化后的图像可采用空气中的图像处理方法, 为空气中的三维重建方法应用于水下的探索提供理论依据.

In order to solve the problem of image distortion of the three-dimensional reconstruction underwater, a model based on the non-parallel binocular stereo vision system was proposed, which could transfer the image shooted through a plat waterproof cover underwater into the image shoted in air, and general conditions were presented to use it. When the focus of lens is on the surface of the water, there is a one-to-one correspondence between the underwater image and the image-without-water. If the focus of lens is outside the water, the proposed model still valid when the distance from the focus to the object is far greater than the focal length and the distance from the focus of lens to refractive surface along z axis. Three-dimensional reconstruction of the feature points was used in experiments to verify the validity of the model. The results show that with the mentioned method, the average coordinate errors of the feature points are 2.23%(x), 1.51%(y), 1.10%(z), respectively, compared with the actual one. The proposed model provides a basis of applying the three-dimensional image reconstruction.