Collaborative Research: CoralReef3D - An open-source toolkit for underwater 3D ecosystem modeling and analysis

珊瑚礁是地球上最多样化的生态系统之一，拥有至少30%的海洋物种。它们每年在全球创造近10万亿美元的收入，为超过5%的人类提供蛋白质和收入。不幸的是，由于气候变化和当地的压力，大多数珊瑚礁正在严重减少。然而，一些珊瑚礁似乎比其他珊瑚礁更有弹性。我们需要迅速找到并识别这些珊瑚礁，以阐明赋予这种恢复力的因素。从历史上看，定位和证明弹性所需的各种调查都是缓慢、昂贵的，并且依赖于训练有素的科学家。这些传统方法过于缓慢和繁琐，无法为我们21世纪的挑战提供21世纪的解决方案。在这个项目中，先进的3D计算机视觉和摄影测量技术将被整合到一个名为CoralReef3D的新型计算工具集中，该工具集将提供给世界各地的潜水员和海岸测量员。这个共享软件将允许非专家创建地理参考（因此可重复）3D珊瑚礁蒙太奇，可以很容易地评估珊瑚生长和生存的基准研究。这项研究的影响将通过开放源代码软件、在线研讨会、学生培训以及涉及博物馆和公众的外展活动来扩大。CoralReef3D有望在水下三维测量方面取得重大进展。在本研究中，将开发新的水折射校正方法，并将其纳入CoralReef3D。对相机/水界面的水折射变形的校正将允许对海景蒙太奇进行精确测量。例如，该程序在物体识别和语义分割方面的先进能力将允许它为线性或体积测量生成图像尺度。这也将使被调查地区的生态系统复杂性（拓扑复杂性）和地貌学的精确计算成为可能。通过为所有图像提供GPS信息，CoralReef3D可以追踪单个群落的命运，并量化生态系统随时间的变化。受控实验室验证和现场验证都将进行。受控的实验室验证将在珊瑚礁系统珊瑚农场维护的深海水池中进行。现场验证将在夏威夷和佛罗里达群岛沿深度梯度进行。为了缩短通常与复杂的图像分析程序相关的学习曲线，CoralReef3D将使用文本和图像说明广泛记录。这些技术进步将为水中复杂物体的三维建模提供一个基准方法。CoralReef3D将使专家和非专业人士能够对快速变化的海洋生态系统进行经济、大规模和准确的评估。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

Coral reefs are one of the most diverse ecosystems on Earth, hosting at least 30% of all marine species. They generate almost $10 trillion USD per year globally and provide protein and income for more than 5% of humanity. Unfortunately, most coral reefs are in severe decline due to the climate change and local stressors. Some reefs, however, appear to be more resilient than others. We need to find and identify these reefs quickly to elucidate the factors that confer this resilience. Historically, the kinds of surveys required to locate and demonstrate resilience have been slow, expensive, and reliant on highly trained scientists. These traditional methods are too slow and too cumbersome to provide 21st Century solutions for our 21st Century challenges. In this project, advanced 3D computer vision and photogrammetric techniques will be assembled into a novel computational toolset called CoralReef3D, which will be made available to divers and coastal surveyors all over the world. This shareware will allow non-experts to create georeferenced (and therefore repeatable) 3D reef montages that can easily be assessed for benchmark studies on coral growth and survival. The impact of this research will be broadened by open-source software, online workshops, student training, and outreach activities involving museums and the general public.CoralReef3D is expected to bring forth significant advancements in underwater 3D measurements. During this research, novel water-refraction correction methods will be developed and incorporated into CoralReef3D. Correction for water-refraction distortions at the camera/water interface will allow production of accurate measurements of seascape montages. For instance, the program’s advanced capabilities for object-recognition and semantic-segmentation will allow it to generate on-image scales for linear or volumetric measurements. This will also enable precise calculations of ecosystem rugosity (topological complexity) and geomorphology in surveyed areas. By providing GPS information on all images, CoralReef3D can fate-track individual colonies and quantify ecosystem changes over time. Both controlled laboratory validation and in situ field validation will be conducted. Controlled laboratory validations will be performed in deep seawater tanks maintained by Reef Systems Coral Farm. In situ field validations will be performed in Hawaii and the Florida Keys along depth gradients. To shorten the learning curve normally associated with sophisticated image analysis programs, CoralReef3D will be extensively documented using both text and pictorial instructions. These technological advances will produce a benchmark method for 3D modeling of complex objects in water. CoralReef3D will empower both experts and non-specialists to make affordable, large-scale, and accurate assessments of rapidly-changing marine ecosystems.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.