基于单视点调节相机模型的水下图像三维重构

水下图像成像过程中,每个场景点所对应的光线在传播过程中均经过多种折射率不同的介质,因此水下图像会产生折射变形,导致传统基于单视点相机模型的水下图像三维重构方法精度极大降低.在理论分析一个新的单视点调节模型对水下图像折射变形的补偿能力的基础上,提出一种高效、自动的水下图像三维重构及其非线性优化方法.为了验证该方法的性能,利用合成数据和真实图像对其进行测试,并依据重构有效性、重构误差等定量指标考察分析了实验结果.实验结果表明,提出的多视图水下三维重构方法可有效补偿水下场景成像空间变形,进而显著提高传统基于单视点相机模型的水下三维重构质量.

Image-based underwater 3D reconstruction with single viewpoint adjustment camera model

In underwater imaging, the light ray associated with each scene point goes through different media with different refractive indices, which introduces refractive distortion in underwater images. Such distortion reduces the accuracy of image-based 3D reconstruction based on traditional single viewpoint camera model. Based on theoretical analysis of the ability of a new single viewpoint adjustment camera model in compensating for refractive distortion, an efficient and automatic image-based underwater 3D reconstruction algorithm with non-linear optimization was proposed. The performance of the purposed method in terms of reconstruction efficiency and accuracy was examined quantitatively using both synthetic data and real images. The experimental results suggest that the proposed single viewpoint adjustment camera model can handle scene space distortion effectively and significantly improve the quality of traditional image-based underwater 3D reconstruction using single viewpoint camera model.