面向作物生长调节的果蔬对靶授粉目标识别与精准控制研究

项目名称： 面向作物生长调节的果蔬对靶授粉目标识别与精准控制研究

项目编号： No.31301232

项目类型： 青年科学基金项目

立项/批准年度： 2014

项目学科： 农业科学

项目作者： 袁挺

作者单位： 中国农业大学

项目金额： 23万元

中文摘要： 果蔬授粉是农作物生产流程中的重要环节。由于设施农业相对密闭空间，授粉以传统人工蘸花为主，劳动强度大、作业效率低、果品易受污染，探索果蔬自动授粉技术符合现代设施农业的发展要求。本项目针对设施栽培作物花簇管理共通性技术，以机器视觉和自动控制为技术手段，面向果蔬生长调节开展自动化对靶授粉目标识别与精准控制研究。提出基于光照色彩恒常特性的花簇图像动态获取方法，建立作物花簇不同波长光的偏振反射模型和光照波动控制模型，突破自然环境光照变化对作物信息动态稳定获取的技术局限。研究基于目标自相似的分形理论和基于光度立体法的准三维图像建模，形成一种新的高效花簇空间信息探测方法，实现多点目标信息实时跟踪与精确定位。探索基于光谱图像与模式识别的花簇生长特征和花粉变量诱导决策关系模型，构建作物花簇生长信息数据库、模型库，利用系统知识规则诊断推理机，实现授粉浓度和剂量的精确调控，为自然环境下自动变量授粉提供依据。

中文关键词： 果蔬授粉；机器视觉；花簇识别；光照补偿；变量调控

英文摘要： Fruit and vegetable pollination is an important link in the process of crop production. Because the space of facility agriculture is confined relatively, the automatic pollination as compared with the traditional artificial dunking flowers would bring labor-intensive, low operating efficiency, and fruit vulnerable to contamination. To develop the automatic pollination technology for fruit and vegetable meets the requirements of the development of the modern agricultural facilities. According to the intercommunity for flower clusters management of facilities cultivated crops, the project uses machine vision and automatic control technology to study the recognition of toward-target pollination and the precision control plan to realize the fruit and vegetable growth regulation. In order to optimize the contrast between lighting environment, background and flower cluster, the reflection and polarizaiton model of different wavelengths for flower cluster will be established. Besides, the image dynamic acquisition system against sunlight fluctuation in greenhouse will be developed which can break the limitation of obtaining the corp information in natural light environments dynamically. The fractal theory based on the target self-similarity and the quasi-three-dimensional image model based on photometric stereo method

英文关键词： Fruit and vegetable pollination；Machine vision；Flower detection；Sunlight compensation；Variable control