Novel computer vision techniques for food quality analysis - identification of Bruchus rufimanus (bean seed beetle) damage in field beans (Vicia faba) for export for human consumption

用于食品质量分析的新型计算机视觉技术 - 识别供人类消费出口的蚕豆（Vicia faba）中的 Bruchus rufimanus（豆籽甲虫）损害

• 批准号: 131442
• 金额: $ 9.26万
• 依托单位: PROCESSORS & GROWERS RESEARCH ORGANISATION
• 依托单位国家: 英国
• 项目类别: Feasibility Studies
• 财政年份: 2013
• 资助国家: 英国
• 起止时间: 2013 至 无数据
• 项目状态: 已结题
• 来源: https://gtr.ukri.org/projects?ref=131442
• 关键词: Novel; computer; vision; techniques; food

The project will aim to analyse field bean produce for human consumption for the the presence of adult Bruchus rufimanus (bean seed beetle) and larval damage. The study will review, test and develop computer state of the art vision algorithms suitable for detecting, selecting and classifying the beans most effectively. A methodology/demonstrator will be proposed to meet the industry's requirements. This will employ computer cision and machine learning techniques in order to automatically analyse images generated from samples from a number of bean crops. These will be classified as good or damaged. Samples will also be analysed by hand using the existing system to calibrate and test the effectiveness of the automated vision based approaches. The study will consider the potential to develop hand-held instruments using the technology that can be employed for rapid analysis of damage and insect presence. Potential uses of the technology for identification of contaminants in food processing, such as poisonous berries and insects will also be considered. The innovative aspects of this project include a fully automatic, robust and accurate system for detecting and classifying samples from digital images, the application of state-of-the-art vision techniques to food technology and development of a prototype system to classify samples from images taken with a simple camera or handset.

该项目的目的是分析供人类消费的田间豆类产品中是否存在Bruchus rufimanus（豆种甲虫）成虫和幼虫损害。这项研究将审查、测试和开发最先进的计算机视觉算法，以最有效地检测、选择和分类咖啡豆。将提出一种方法/示范，以满足行业的要求。这将采用计算机精确和机器学习技术，以便自动分析从一些豆类作物的样本中生成的图像。这些将被分类为良好或损坏。此外，亦会使用现有系统手动分析样本，以校正及测试自动视觉方法的成效。这项研究将考虑利用可用于快速分析损害和昆虫存在的技术开发手持仪器的潜力。还将考虑该技术在食品加工中识别污染物（如有毒浆果和昆虫）的潜在用途。该项目的创新方面包括一个全自动，强大和准确的系统，用于从数字图像中检测和分类样品，将最先进的视觉技术应用于食品技术，以及开发一个原型系统，用于从简单的相机或手机拍摄的图像中分类样品。