基于机器视觉的不结球白菜幼苗光响应机制的研究

Using non-destructive monitoring of crop interception of photosynthetically active radiation (PAR) to establish the correlation of crop growth status, leaf area index and photosynthetic capacity, and to precisely characterize the dynamics of crop growth phenotype, physiological state and photosynthesis utilization, could provide new strategy for protected environment illumination regulation and precise operation. This project takes pakchoi (Brassica campestris L.) seedling as the research object, and for the problems of long detection time, instability, and high cost of photosynthesis measurement, aims to develop digital image-based light interception acquisition model to reveal the real-time response mechanism of seedling to the illumination change. This project will also investigate the correlation of light interception with photosynthetic rate and leaf area index, and compare seedling PAR conversion rate and utilization rate under different seedling densities and growth stages, which will provide technical support for illumination precise regulation and fertigation decision making in industrialized seedling production.

通过无损监测作物对光合有效辐射的截获率，建立作物生育状态、叶面积指数及光合能力的相关关系模型，精确表征作物生长表型、生理状态与光合利用率等动态过程，能够为设施环境的光照调控、精准作业提供新的方法。本项目以不结球白菜幼苗为研究对象，针对光合作用等生理生态检测时间长、不稳定、成本高等问题，通过建立基于数字图像的光截获率检测模型，揭示幼苗对光照变化的实时响应机制；通过研究光截获率与光合速率、叶面积指数的相关性，研究比较不同育苗密度、生育时期幼苗光合有效辐射转化率和利用率，为工厂化育苗光环境精准调控和水肥决策提供技术参考。

项目通过无损监测作物表型特征，建立了不结球白菜幼苗生育状态、光和有效辐射截获率的相关关系模型，精确表征幼苗生育动态变化规律，形成基于机器视觉的不结球白菜幼苗栽培管理方法。（1）针对穴盘栽培和潮汐灌溉模式，研发了用于苗床的无线控制的自走式信息采集系统，可在水平及垂直平面上进行快速无损的植株信息采集；（2）研究了不结球白菜幼苗数字图像与生理指标的相关性。筛选了显著相关的图像特征参数，构建了基于数字图像的幼苗干重、鲜重、叶绿素和叶面积指数检测模型。（3）建立了不结球白菜幼苗光截获率检测模型。筛选了与光截获率实测值显著相关的图像特征参数，对每个特征参数进行回归拟合，得到光截获率预测模型（R2=0.9)。（4）解析了不同规格穴盘幼苗生长动态规律并进行品质综合评价。针对生长动态变化，结合表型指标和生理指标，解析了不结球白菜幼苗生长品质动态变化规律。针对指标的表现差异，基于主成分分析法，建立了幼苗品质综合评价指数。（5）建立了基于机器视觉的不结球白菜幼苗栽培管理方法。选取水分这一重要栽培要素，利用表型分析和机器学习建模方法，构建了植株根域水分状态判别模型。综合上述结果，本研究为精准育苗管理和水肥决策提供了参考。