Advanced machine learning models and methods for hyperspectral imagery processing and analysis

用于高光谱图像处理和分析的先进机器学习模型和方法

• 批准号: RGPIN-2019-06744
• 负责人: Xu, Linlin
• 金额: $ 2.26万
• 依托单位: University of Waterloo
• 依托单位国家: 加拿大
• 项目类别: Discovery Grants Program - Individual
• 财政年份: 2021
• 资助国家: 加拿大
• 起止时间: 2021-01-01 至 2022-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=739083
• 关键词: Advanced; machine; learning; models; methods

Hyperspectral imaging, which records the reflected radiation from the earth surface using hundreds of narrow and contiguous spectral image bands, can provide rich information for identifying and distinguishing spectrally similar materials, and has become an essential tool leveraged by government agencies, research institutions and commercial companies to discern materials and map surface properties in various resource and environment applications. Nevertheless, due to the large data volume, the innate high-dimensionality, spatial-spectral heterogeneities and the noise effect of hyperspectral image (HSI), there are significant challenges for operational users and research scientists to efficiently and accurately transform HSI into value-added information. Therefore, novel models and methods for computer-aided processing and analysis of the large-volume complex HSI can significantly improve the quantification of valuable geochemical, biochemical and biophysical variables, and thereby improve natural resource management and environment monitoring. The proposed research program aims to develop novel intelligent models and computational methods for hyperspectral image processing and analysis. The following objectives will be investigated: 1) a novel spectral unmixing based HSI representation framework with strong modeling capacity and efficient optimization approach, capable of fully addressing various key characteristics of HSI, and disentangling the underlying explanatory factors in HSI for supporting the development of enhanced task-specific algorithms, 2) highly accurate and efficient algorithms tailored to different HSI processing tasks (e.g., denoising, super-resolution, feature extraction, classification, unmixing, end-member extraction, sub-pixel mapping) to enable full exploration of the potential of HSI, 3) advanced HSI software tools leveraging the optimized image processing algorithms to facilitate the development of data processing workflows, pipelines and analytic solutions to key HSI applications for mining value-added information in HSI. The proposed research program will improve the current state-of-the-arts in HSI modeling theories, algorithms and software tools, and thus advance new capabilities in environment monitoring, assessment and protection, as well as natural resource management and exploration. Moreover, HQP will be trained in hyperspectral image processing and analysis, remote sensing, computer vision and artificial intelligence within a multidisciplinary environment, making them in high demand in the billion dollar job market, and enabling them to find relevant, leadership opportunities in Canadian industry, government and academia.

高光谱成像利用数百个狭窄而连续的光谱图像波段记录来自地球表面的反射辐射，可以为识别和区分光谱相似的物质提供丰富的信息，并且已经成为政府机构、研究机构和商业公司在各种资源和环境应用中识别物质和绘制表面特性的重要工具。然而，由于高光谱图像（HSI）的数据量大、固有的高维性、空间-光谱异质性和噪声效应，如何有效、准确地将HSI转化为增值信息对业务用户和研究人员来说是一个重大挑战。因此，用于大体积复杂HSI的计算机辅助处理和分析的新模型和方法可以显著改善有价值的地球化学、生物化学和生物物理变量的量化，从而改善自然资源管理和环境监测。 拟议的研究计划旨在开发新的智能模型和高光谱图像处理和分析的计算方法。将研究以下目标：1）一种新颖的基于光谱解混的HSI表示框架，其具有强大的建模能力和有效的优化方法，能够完全解决HSI的各种关键特性，并解开HSI中的潜在解释因素，以支持增强的任务特定算法的开发，2）针对不同HSI处理任务（例如，去噪、超分辨率、特征提取、分类、解混、端元提取、亚像素映射），以充分挖掘HSI的潜力，3）先进的HSI软件工具，利用优化的图像处理算法，促进数据处理工作流程、管道和分析解决方案的开发，以挖掘HSI中的增值信息。拟议的研究计划将改善HSI建模理论，算法和软件工具的当前最先进水平，从而提高环境监测，评估和保护以及自然资源管理和勘探的新能力。此外，HQP将在多学科环境中接受高光谱图像处理和分析，遥感，计算机视觉和人工智能方面的培训，使他们在数十亿美元的就业市场中需求旺盛，并使他们能够在加拿大工业，政府和学术界找到相关的领导机会。