ABI Development: Ecosynth: An Advanced Open-Source 3D Toolkit for Forest Ecology

马里兰大学巴尔的摩县分校获得了一笔拨款，用于开发Ecosynth，这是一个基于计算机视觉技术最新创新的开源3D工具包，用于扫描林地生态系统，并与一个在线社区系统相结合，用于浏览、共享、可视化、标记和分析陆地生态系统的3D扫描。三维扫描技术使对林地（森林和灌丛）生态系统结构和功能的科学理解取得了根本性进展，同时为快速评估景观生物多样性和碳储量提供了新的工具。现有的3D扫描技术都是基于精密激光扫描系统，其成本非常昂贵，而且在技术上具有挑战性，限制了其应用于相对较少的现场、研究人员和时间段。该项目旨在通过使生态学家和公民科学家能够常规和频繁地获取、使用和共享3D扫描数据，从而改变林地野外生态学的实践。这将使遥感和先进的3D分析成为“用户驱动”而不是“专家驱动”的技术。Ecosynth 3D扫描技术将Structure from Motion算法应用于使用普通消费级数码相机（包括配备相机的手机）在现场获取的图像，这些相机可以在地面或由业余爱好者遥控飞机以计算机优化的模式部署。由该项目开发的Ecosynth工具包将作为开源开发资源在社区网站（http://ecotope.org/projects/ecosynth/）上提供，并作为一组工具提供给潜在用户，从而实现以下功能：1)计算机辅助3D扫描（地面和空中）的最佳图像采集，2)快速生成高空间分辨率地理参考3D扫描（多光谱点云），3)基于浏览器的在线系统，用于共享和可视化陆地生态系统的3D扫描，使专家和其他用户能够识别、标记和在3D虚拟环境中对生态对象进行其他空间明确标记。这些能力的发展和扩展将把生态观察和信息学的前沿带到最有实地经验的人手中，有可能使景观生态学的“群众外包”成为可能。林地目前是REDD+和其他国际协议的碳和生物多样性保护目标，因此是扩大科学研究和监测的关键领域——Ecosynth工具将有助于这些努力。让野外生态学家和民间科学家定期廉价地使用3D成像工具，将使森林观测能力扩大到更广泛的观察者群体，从而有助于全球森林保护工作，甚至城市林业。该项目是建立在跨学科团队的基础上的，学生在其职业生涯的多个阶段，从高中生（实地测试），本科生（生态学，地理学，数学，计算机科学，电气工程，机械工程和信息系统），研究生（地理与环境系统，计算机科学与工程）和博士后研究人员。工作还将吸引公众，由一个顾问小组指导，该小组包括学科专家（生态学、计算机科学）和公民科学项目的保护组织。最终，Ecosynth项目旨在激发一个开源3D生态社区的发展，致力于开发、共享和调查基于使用计算机视觉的3D扫描结果的林地生态。

The University of Maryland Baltimore County is awarded a grant to develop Ecosynth, an open-source 3D toolkit for scanning woodland ecosystems based on recent innovations in computer-vision technologies coupled with an online community system for browsing, sharing, visualizing, tagging and analyzing 3D scans of terrestrial ecosystems. Three dimensional (3D) scanning technologies are enabling fundamental advances in scientific understanding of ecosystem structure and function in woodlands (forests and shrublands) while providing new tools for rapid assessment of biodiversity and carbon storage across landscapes. Existing 3D scanning technologies are based on precision laser scanning systems, which remain prohibitively expensive and technically challenging to deploy, limiting their application to a relatively small number of field sites, investigators and time periods. This project aims to transform the practice of field ecology in woodlands by enabling the routine and frequent acquisition, use, and sharing of 3D scanning data by both ecologists and citizen scientists. This will enable remote sensing and advanced 3D analytics to become a "user-driven" rather than "expert-driven" technology. Ecosynth 3D scanning technology applies Structure from Motion algorithms to images acquired in the field using ordinary consumer-grade digital cameras, including camera-equipped cell phones, deployed in computer-optimized patterns on the ground or from the air by hobbyist remote controlled aircraft. The Ecosynth toolkit developed by this project will be offered as an open-source development resource on the community website (http://ecotope.org/projects/ecosynth/), and to potential users as a set of tools enabling the following capabilities: 1) computer assisted optimal image acquisition for 3D scanning (ground and aerial), 2) rapid generation of high-spatial resolution georeferenced 3D scans (multispectral point clouds), 3) an online browser-based system for sharing and visualizing 3D scans of terrestrial ecosystems that will enable expert and other users to identify, tag and make other spatially explicit markups of ecological objects in a 3D virtual environment. Development and extension of these capabilities will bring the cutting edge of ecological observations and informatics into the hands of those with the most field experience, potentially enabling the "crowd-sourcing" of landscape ecology.Woodlands are currently targeted for carbon and biodiversity conservation by REDD+ and other international agreements and are therefore key areas for expanded scientific research and monitoring- Ecosynth tools will aid in these efforts. Enabling regular inexpensive use of 3D imaging tools by field ecologists and citizen scientists will expand forest observation capabilities to a much wider pool of observers, thereby assisting in global forest conservation efforts and even urban forestry. The project is built upon engaging students in interdisciplinary teams across multiple stages in their careers, from high school students (field testing), undergraduates (Ecology, Geography, Math, Computer Science, Electrical Engineering, Mechanical Engineering, and Information Systems), graduate students (Geography & Environmental Systems, Computer Science & Engineering), and postdoctoral researchers. Work will also engage the public, guided by an advisory panel including both disciplinary experts (ecology, computer science) and conservation organizations with citizen science programs. Ultimately the Ecosynth project aims to spark the development of an open source 3D ecology community dedicated to developing, sharing and investigating the ecology of woodlands based on the results of 3D scanning using computer vision.