SBIR Phase II: High-Resolution Image Segmentation for Natural Resource Management

这一小型企业创新研究(SBIR)第二阶段项目的商业/更广泛影响是为美国提供经济效益、健康优势和改进的自然灾害应对准备。从历史上看，自然资源和保护组织在绘制其目标生态系统地图方面遇到了困难，无论是由于人工调查的高成本，还是由于成像技术的低分辨率。每年，各组织在地理空间监测和分析上的花费超过270亿美元。该第二阶段项目将在提高分辨率的同时降低生态系统测绘的成本，从而实现现有的最高质量的植被健康跟踪。此外，该项目将导致自然资源测绘的工作时间减少50%-90%。通过节省时间，利益相关者可以将精力分配到自然资源管理的其他方面。通过绘制土地使用随时间变化的地图，管理者和自然资源保护者可以跟踪土地变化，并确定目前实施的项目是否对生态系统产生了预期的影响。该项目还将改善对整个流域的植被的监测和管理，这些流域提供了美国大约80%的饮用水系统，这些系统的水质依赖于健康和生物多样性的植被来过滤污染物。最后，这个项目将提高政府机构快速监测自然灾害对环境的影响并通知响应的能力。这个小企业创新研究第二阶段项目将开发一个全面的软件系统，可以在不同的景观场景中提供无与伦比的光谱和空间细节。与目前的劳动密集型现场测试相比，该项目的输出将提供相当或更好的细节水平的场景表征，同时克服了历史上排除全面和重复监测的时间、成本和可获得性限制。这些第二阶段目标的实现将产生一个用户友好的土地复盖测绘系统，该系统将能够进行高分辨率的环境监测。关于人口动态、气候变化引起的植被变化和疾病评估的系统输出可以促进精确的生态系统管理和气候行动的数据驱动决策。创新的框架包括三个主要部分：1)图像的预处理和变换，2)图像分割，3)分辨率恢复。这种方法提供了生态系统类型之间的快速可复制性和由于处理效率而具有的通用性可扩展性，同时提供了目前无法获得的生态系统健康指标。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

The commercial/broader impact of this Small Business Innovation Research (SBIR) Phase II project is to provide economic benefits, health advantages, and improved natural disaster response readiness to the USA. Historically, natural resource and conservation organizations have had difficulties mapping their targeted ecosystems, whether due to high costs of manual surveys or poor resolution of imaging technologies. Annually, organizations spend more than $27 billion on geospatial monitoring and analysis. This Phase II project will decrease the cost of ecosystem mapping while increasing resolution, allowing for the best quality vegetation health tracking available. Additionally, this project will result in a 50-90% reduction in work hours for natural resource mapping. By saving time, stakeholders can allocate effort to other aspects of natural resource management. By mapping land use over time, managers and conservationists can track land changes and determine if currently-implemented programs are having intended impacts on the ecosystem. This project will also improve monitoring and managing of vegetation across watersheds that provide roughly 80% of US drinking water - systems where water quality relies on healthy and biodiverse vegetation to filter pollutants. Lastly, this project will improve the ability of government agencies to rapidly monitor environmental impacts of natural disasters and inform responses.This Small Business Innovation Research Phase II project will develop a comprehensive software system that can provide unparalleled spectral and spatial detail on diverse landscape scenes. Compared to current labor-intensive field testing, this project’s outputs will offer scene characterization at comparable, or better, levels of detail, while surmounting the time, cost, and accessibility constraints that have historically precluded comprehensive and repetitive monitoring. Accomplishment of these Phase II goals will yield a user-friendly land cover mapping system that will enable high-resolution environmental monitoring. System outputs on population dynamics, climate change-induced vegetation shifts, and disease assessments can facilitate data-driven decision-making for precision ecosystem management and climate action. The framework of the innovation consists of three main components: 1) image pre-processing and alteration, 2) image segmentation, and 3) resolution recovery. This approach provides rapid replicability between ecosystem types and versatile scalability due to processing efficiency, while providing currently unavailable ecosystem health indicators.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.