基于多约束的水下小目标高精度三维重建研究

项目名称： 基于多约束的水下小目标高精度三维重建研究

项目编号： No.61501278

项目类型： 青年科学基金项目

立项/批准年度： 2016

项目学科： 无线电电子学、电信技术

项目作者： 吴则举

作者单位： 青岛理工大学

项目金额： 20万元

中文摘要： 水下声学成像技术主要采用水下声纳系统，声波波长较长，绕射能力强，受水质影响小，探测范围大，但其无法获取水下目标真实的光学信息，水平分辨率和竖直分辨率较低，即使是当前世界上先进的3D成像声纳，其竖直分辨率也只能达到4cm，水平分辨率达8mm，难以观测水下表面纵向高度差4cm以下小目标。针对竖直高度差在4cm以内的水下小目标高精度三维重建问题，本项目研究，利用水下朗伯模型和水下光度立体学，拟提出一种水下光学成像模型参数估计算法，以降低水下吸收和散射的影响；在此基础上，提出非线性最优的水下小目标物体表面法向和光学反照率求解算法；基于结构光特征点的水下小目标表面高度重建约束，水下小目标表面法向约束和水下小目标表面平滑约束，实现基于多约束的水下小目标高精度三维重建目标。项目研究成果可广泛应用于人工鱼礁微生物附着检测，海洋及水下工程监测、水下装备高精度监测、水下考古及海底资源开发等方面。

中文关键词： 水下高度重建；水下表面法向重建；水下光学三维重建

英文摘要： Techniques for underwater three-dimensional reconstruction can be classified into two types, i.e. those based on acoustic imaging and optical imaging respectively. Underwater sonar systems are adopted for acoustic imaging; they use relatively long wavelength, and have strong diffraction ability, less influence from water quality and broad detection range. However, they suffer from the unavailability to acquire real optical information of underwater objects, as well as the relatively low horizontal and vertical resolution. Even with the most advanced 3D imaging sonar, the vertical resolution can hardly reach to 4cm, while the horizontal resolution can be up to 8mm. It is difficult to observe surface details of small objects with surface height difference below 4cm. .we investigate 3D reconstruction of underwater small target with a height difference less than 4cm. In this project, By integrating the Lambertian model and underwater photometric stereo, we propose a new algorithm for estimating parameters of underwater optical transmission model which reduces the effects of absorption and scattering of underwater. Based on this model, we propose an algorithm for finding a non-linear optimal solution to the normal direction and the reflection coefficient of the underwater object surface. With the constraint of the height reconstruction results on the structured-light feature points on the object surface, and by using the constraints from the reconstructed surface normal and the constraints from the surface smoothness of underwater small object, we propose a new algorithm for high-precision reconstruction of underwater object surface height. The proposed method can be widely used in detection of microbial attachment of artificial reef, high-precision monitoring of underwater equipment, underwater archeology and seabed resources development.

英文关键词： Underwater height reconstruction;Underwater surface normal;Underwater optical three-dimensional