16AGRITECHCAT5: Feasibility of a Hyper Spectral Crop Camera (HCC) for agriculture optimisation

16AGRITECHCAT5：用于农业优化的高光谱作物相机 (HCC) 的可行性

• 批准号: BB/P004873/1
• 负责人: Stephen Marshall
• 金额: $ 8.66万
• 依托单位: University of Strathclyde
• 依托单位国家: 英国
• 项目类别: Research Grant
• 财政年份: 2016
• 资助国家: 英国
• 起止时间: 2016 至 无数据
• 项目状态: 已结题
• 来源: https://gtr.ukri.org/projects?ref=BB%2FP004873%2F1
• 关键词: 16AGRITECHCAT5; Feasibility; Hyper; Spectral; Crop

Farmers and horticulturists face varying difficulties that require experience and knowledge of their fields and crops, gained over many years. These difficulties include, but are not limited to: uneven growth/yield of their fields; inexact and estimated fertiliser application; uneven irrigation and local variations in pests/diseases/weeds. Additionally, the optimum harvest timing is still speculated and often inexact. Faced with numerous variables, farmers cannot avoid high variations in costs and crop yields from year to year. Tools to assist farmers to optimise e.g. fertiliser & water applications or early detection of disease will provide a useful diagnostic and management capability for optimum control of crop growth. Currently, solutions for these challenges do exist, however, current systems are large, heavy, not portable and as such are not readily deployable. They are also prohibitively expensive - typically £10,000 - £150,000 each - and are generally only suitable for use in airborne or satellite imaging applications or laboratory analysis. In effect, the current solutions available for the aforementioned agricultural challenges are limited to large scale farming and/ or high value crops. In these expensive systems, a spectrometer scan or image of the crop is taken at visible and/or infrared wavelengths with analysis showing spectral image signature changes relating to crop growth conditions. The signatures of interest varies from plant to plant and from cause to cause. The "colour" of a crop (visible and IR) also changes as it approaches maturity, with spectrometer scans providing scientific information for informed management decisions in relation to crop hydration, fertiliser application, disease progression and harvesting. Hyperspectral Imaging (HSI) can capture these changes: HSI systems capture a large number of images of the scene, each at a different wavelength within some range determined by the sensor technology, to produce a so called hyperspectral data cube in which each pixel in the spatial domain contains a spectral profile of the object observed. For our application, this spectral information can be analysed to make decisions about the diagnostics/management of challenges in maximising crop yield. The proposed Hyperspectral Crop Camera (HCC) will be: low-cost, compact & portable, simple in operation and robust. A camera housing will contain the sensor, battery and electronics to produce one small simple lightweight device. This device would be suitable for handheld use or potentially mountable in a low cost drone for local airborne analysis. HSI technology in farming and agriculture which can cost anything from £10k - £150k. Application of HCC can allow a farmer and/ or agriculturists to: - Save water by providing optimised or localised irrigation - Timely identify areas of pests/diseases/weeds for early intervention - Optimise use of fertiliser - Determine optimum harvest time and help increase crop yield - Improve evenness of crop yield across field area - Reduced man hours, manually surveying fields etc - Reduce need for technical agronomy training/knowledge.

农民和园艺家面临着各种各样的困难，需要多年来积累的关于他们的田地和作物的经验和知识。这些困难包括但不限于：田地的生长/产量不均衡;施肥不准确和估计;灌溉不均衡和害虫/疾病/杂草的地方差异。此外，最佳收获时间仍然是推测的，而且往往不准确。面对众多的变量，农民无法避免每年成本和作物产量的巨大变化。帮助农民优化肥料和水的施用或早期发现疾病的工具将为作物生长的最佳控制提供有用的诊断和管理能力。目前，针对这些挑战的解决方案确实存在，但是，当前的系统体积大、重量重、不便携，因此不容易部署。它们也非常昂贵-通常每个10，000 - 150，000英镑-并且通常仅适用于机载或卫星成像应用或实验室分析。实际上，目前可用于上述农业挑战的解决方案仅限于大规模农业和/或高价值作物。在这些昂贵的系统中，在可见光和/或红外波长下拍摄农作物的光谱仪扫描或图像，分析显示与农作物生长条件相关的光谱图像特征变化。所关注的特征因植物和原因而异。作物的“颜色”（可见光和红外线）也会随着作物接近成熟而发生变化，光谱仪扫描可提供科学信息，以便在作物水合作用、化肥施用、疾病进展和收获方面做出明智的管理决策。高光谱成像（HSI）可以捕捉到这些变化：HSI系统捕获大量的场景图像，每个图像都在传感器技术确定的某个范围内的不同波长处，以产生所谓的高光谱数据立方体，其中空间域中的每个像素都包含所观察对象的光谱轮廓。对于我们的应用，可以分析该光谱信息以做出关于最大化作物产量的挑战的诊断/管理的决策。建议的高光谱作物相机（HCC）将是：低成本，紧凑和便携式，操作简单，坚固耐用。相机外壳将包含传感器，电池和电子产品，以生产一个小型简单的轻量级设备。该设备将适合于手持使用或潜在地安装在低成本无人机中用于本地空中分析。HSI技术在农业和农业，可以花费任何从£ 10 k- £ 150 k。HCC的应用可以允许农民和/或农学家：- 通过提供优化或局部灌溉来节约用水-及时识别害虫/疾病/杂草的区域以进行早期干预-优化肥料的使用-确定最佳收获时间并帮助提高作物产量-提高整个田地区域的作物产量均匀性-减少工时，人工测量田地等-减少对技术农艺培训/知识的需求。

项目成果

Near-infrared hyperspectral imaging for non-destructive classification of commercial tea products

• DOI: 10.1016/j.jfoodeng.2018.06.015
• 发表时间: 2018-12-01
• 期刊: JOURNAL OF FOOD ENGINEERING
• 影响因子: 5.5
• 作者: Mishra, Puneet;Nordon, Alison;Marshall, Stephen
• 通讯作者: Marshall, Stephen

Varietal Classification of Rice Seeds Using RGB and Hyperspectral Images

• DOI: 10.1109/access.2020.2969847
• 发表时间: 2020-01-01
• 期刊: IEEE ACCESS
• 影响因子: 3.9
• 作者: Fabiyi, Samson Damilola;Vu, Hai;Marshall, Stephen
• 通讯作者: Marshall, Stephen

Superpixel based Feature Specific Sparse Representation for Spectral-Spatial Classification of Hyperspectral Images

高光谱图像光谱空间分类的基于超像素的特征特定稀疏表示

• DOI: 10.3390/rs11050536
• 发表时间: 2019-03-01
• 期刊: REMOTE SENSING
• 影响因子: 5
• 作者: Sun, He;Ren, Jinchang;Marshall, Stephen
• 通讯作者: Marshall, Stephen

Removal of specular reflections from image sequences using feature correspondences

• DOI: 10.1007/s00138-017-0826-6
• 发表时间: 2017-02
• 期刊: Machine Vision and Applications
• 影响因子: 3.3
• 作者: S. M. Z. A. Shah;S. Marshall;P. Murray