SBIR Phase II: Immediate Delivery of Massive Aerial Imagery to Farmers and Crop Consultants

小企业创新研究（SBIR）第二阶段项目将加速采用数据密集型精准农业，提高产量，同时减少化肥和农药等农业投入。该项目消除了处理无人机系统进行的大型航空调查的软件瓶颈（时间和劳动力），从而实现了向农民提供可操作信息的成本效益和及时流程。使用频繁的高质量航空扫描，农民可以优化肥料的使用，并更精细地控制增加作物产量所需的杀虫剂和除草剂的数量。此外，农民能够发现问题区域，最大限度地减少植物疾病的传播，并确定诸如积水，灌溉故障和种植或耕作中持续的自动化机械错误等问题，从而降低成本和损失。该项目为互联网基础设施不足的农村客户提供了特殊的好处，因为他们不需要将大量图像上传到云端进行处理。这项技术是农业领域广泛倡议的一部分，该倡议旨在解决到2050年大幅增加粮食产量的需求，以应对世界的预期增长。该项目将继续开发用于空中图像拼接的实时正射校正、拼接和标准化的算法，并将其部署在易于使用的软件原型中。第一阶段已经展示了这种图像处理的行业领先速度。该研究项目背后的技术是从头开始设计的，相对于其他方法，它可以用更少的内存和更快的速度处理大量数据，并通过专有的流图像表示实现，允许将多通道千兆像素和兆兆像素图像视为普通图像。第二阶段支持对软件进行新的扩展，以简化和加速提供拼接和分析的地图，例如优先考虑客户正在探索的图像区域的计算。这将有效地消除无人驾驶航空器上的图像获取与何时可以使用之间的延迟。作物顾问已经将其确定为一种变革能力，因为它可以在同一次实地考察中从航空图像中获得地面实况信息，从而节省时间和劳动力。该项目支持的计算有限环境中的性能增益是收集信息的新能力与农民之间的关键联系。的能力，利用它来提高生产力，同时降低成本。

This Small Business Innovation Research (SBIR) Phase II project will accelerate the adoption of data intensive precision agriculture, increasing yields while decreasing farm inputs such as fertilizers and pesticides. This project removes the software bottleneck (time and labor) in processing large aerial surveys taken by Unmanned Aerial Systems, enabling a cost-effective and timely process to deliver actionable information to farmers. Using frequent high-quality aerial scans, farmers may optimize the use of fertilizers and more finely control the amount of pesticides and herbicides necessary to increase crop yield. Furthermore, farmers mitigate costs and losses by being able to spot problem areas, minimize the spread of plant diseases, and identify issues such as standing water, irrigation malfunctions, and persistent automated machinery errors in planting or cultivation. This project provides special benefit for rural customers having inadequate internet infrastructure by eliminating the need to upload massive imagery to the cloud for processing. The technology is part of a broad initiative in agriculture addressing the need for large increases in food production by 2050 in response to the projected growth of the world?s population to over 9 Billion people.This project will continue development of algorithms for on-the-fly orthorectification, stitching, and normalization of aerial image mosaics and their deployment in an easy-to-use software prototype. The Phase I already demonstrated industry-leading speeds for such image processing. The technology behind this research project is designed from the ground up to process massive data with less memory and increased speed relative to other approaches, enabled by a proprietary streaming image representation, that allows multichannel gigapixel and terapixel images to be treated as ordinary images. This Phase II supports new extensions to the software that simplify and accelerate delivering a stitched and analyzed map, such as prioritizing computation in regions of the image that a customer is exploring. This would effectively eliminate the delay between image acquisition on unmanned aerial vehicles and when it can be used. Crop consultants have identified this as a transformative capability, as it enables ground-truthing information derived from aerial imagery in the same field visit, saving time and labor. The performance gains in compute-limited environments supported by this project are a key link between new capabilities to gather information and a farmer?s ability to utilize it to increase productivity while reducing costs.