Taking 65 cherry tomato core germplasms as experimental materials,the genetic diversity of seven agronomic traits were analyzed.The correlation between any two of the seven agronomic traits and the genetic relationshi...Taking 65 cherry tomato core germplasms as experimental materials,the genetic diversity of seven agronomic traits were analyzed.The correlation between any two of the seven agronomic traits and the genetic relationships of these germplasms were analyzed based on genotypic values.The genetic diversity indices of the seven agronomic traits were 4.15,4.13,4.16,4.13,4.13,4.13 and 4.01,respectively,showing that the cherry tomato core collection had abundant genetic diversity.The correlation analysis between traits based on genotype effect values showed that leaf length was significantly correlated with leaf width with the correlation coefficient of 0.56.The fruit width was significantly correlated with fruit length with the correlation coefficient of 0.52.The flesh thickness was significantly correlated with fruit length and fruit width with the correlation coefficients of 0.49 and 0.39,respectively.The single fruit weight was significantly correlated with fruit length,fruit width and flesh thickness with the correlation coefficients of 0.44,0.61 and 0.62,respectively.When the genetic distances between core germplasms of cherry tomato were calculated based on the phenotypic values,65 core germplasms of cherry tomato were divided into three groups with the rescaled distance of 10.When the genetic distances between core germplasms of cherry tomato were calculated based on the genotypic values,the 65 core germplasms of cherry tomato were divided into four groups with the rescaled distance of 10.Comparing the 2 clustering results,it could find that genotypic value-based clustering analysis could better clarify the genetic relationship between core germplasms.This study could provide a theoretical basis for the effective utilization of cherry tomato core germplasms.展开更多
准确识别定位采摘点,根据果梗方向,确定合适的采摘姿态,是机器人实现高效、无损采摘的关键。由于番茄串的采摘背景复杂,果实颜色、形状各异,果梗姿态多样,叶子藤枝干扰等因素,降低了采摘点识别准确率和采摘成功率。针对这个问题,考虑番...准确识别定位采摘点,根据果梗方向,确定合适的采摘姿态,是机器人实现高效、无损采摘的关键。由于番茄串的采摘背景复杂,果实颜色、形状各异,果梗姿态多样,叶子藤枝干扰等因素,降低了采摘点识别准确率和采摘成功率。针对这个问题,考虑番茄串生长特性,提出基于实例分割的番茄串视觉定位与采摘姿态估算方法。首先基于YOLACT实例分割算法的实例特征标准化和掩膜评分机制,保证番茄串和果梗感兴趣区域(Region of interest,ROI)、掩膜质量和可靠性,实现果梗粗分割;通过果梗掩膜信息和ROI位置关系匹配可采摘果梗,基于细化算法、膨胀操作和果梗形态特征实现果梗精细分割;再通过果梗深度信息填补法与深度信息融合,精确定位采摘点坐标。然后利用果梗几何特征、八邻域端点检测算法识别果梗关键点预测果梗姿态,并根据果梗姿态确定适合采摘的末端执行器姿态,引导机械臂完成采摘。研究和大量现场试验结果表明,提出的方法在复杂采摘环境中具有较高的定位精度和稳定性,对4个品种的番茄串采摘点平均识别成功率为98.07%,图像分辨率为1280像素×720像素时算法处理速率达到21 f/s,采摘点图像坐标最大定位误差为3像素,深度误差±4 mm,成功定位采摘点后采摘成功率为98.15%。与现有的同类方法相比,采摘点图像坐标定位精度提高76.80个百分点,采摘成功率提高15.17个百分点,采摘效率提高31.18个百分点,满足非结构化种植环境中番茄串采摘需求。展开更多
采摘点的识别与定位是智能采摘的关键技术,也是实现高效、适时、无损采摘的重要保证。针对复杂背景下番茄串采摘点识别定位问题,提出基于RGB-D信息融合和目标检测的番茄串采摘点识别定位方法。通过YOLOv4目标检测算法和番茄串与对应果...采摘点的识别与定位是智能采摘的关键技术,也是实现高效、适时、无损采摘的重要保证。针对复杂背景下番茄串采摘点识别定位问题,提出基于RGB-D信息融合和目标检测的番茄串采摘点识别定位方法。通过YOLOv4目标检测算法和番茄串与对应果梗的连通关系,快速识别番茄串和可采摘果梗的感兴趣区域(Region of Interest,ROI);融合RGB-D图像中的深度信息和颜色特征识别采摘点,通过深度分割算法、形态学操作、K-means聚类算法和细化算法提取果梗图像,得到采摘点的图像坐标;匹配果梗深度图和彩色图信息,得到采摘点在相机坐标系下的精确坐标;引导机器人完成采摘任务。研究和大量现场试验结果表明,该方法可在复杂近色背景下,实现番茄串采摘点识别定位,单帧图像平均识别时间为54 ms,采摘点识别成功率为93.83%,采摘点深度误差±3 mm,满足自动采摘实时性要求。展开更多
基金Supported by Hainan Provincial Science and Technology Project(ZDYF2018035)
文摘Taking 65 cherry tomato core germplasms as experimental materials,the genetic diversity of seven agronomic traits were analyzed.The correlation between any two of the seven agronomic traits and the genetic relationships of these germplasms were analyzed based on genotypic values.The genetic diversity indices of the seven agronomic traits were 4.15,4.13,4.16,4.13,4.13,4.13 and 4.01,respectively,showing that the cherry tomato core collection had abundant genetic diversity.The correlation analysis between traits based on genotype effect values showed that leaf length was significantly correlated with leaf width with the correlation coefficient of 0.56.The fruit width was significantly correlated with fruit length with the correlation coefficient of 0.52.The flesh thickness was significantly correlated with fruit length and fruit width with the correlation coefficients of 0.49 and 0.39,respectively.The single fruit weight was significantly correlated with fruit length,fruit width and flesh thickness with the correlation coefficients of 0.44,0.61 and 0.62,respectively.When the genetic distances between core germplasms of cherry tomato were calculated based on the phenotypic values,65 core germplasms of cherry tomato were divided into three groups with the rescaled distance of 10.When the genetic distances between core germplasms of cherry tomato were calculated based on the genotypic values,the 65 core germplasms of cherry tomato were divided into four groups with the rescaled distance of 10.Comparing the 2 clustering results,it could find that genotypic value-based clustering analysis could better clarify the genetic relationship between core germplasms.This study could provide a theoretical basis for the effective utilization of cherry tomato core germplasms.
基金The project supported by National Center of Technology Innovation for Comprehensive Utilization of Saline-Alkali Land(GYJ2023004)国家重点研发计划项目-地理标志产品特色品质控制技术研究与应用(2022YFF0606800)+2 种基金北京市自然科学基金资助(6242028)国家现代农业技术产业体系建设专项基金项目(CARS-23-E03)中央级公益性科研院所基本科研业务费专项(IVF-BRF2024013、Y2023LM10)。
文摘准确识别定位采摘点,根据果梗方向,确定合适的采摘姿态,是机器人实现高效、无损采摘的关键。由于番茄串的采摘背景复杂,果实颜色、形状各异,果梗姿态多样,叶子藤枝干扰等因素,降低了采摘点识别准确率和采摘成功率。针对这个问题,考虑番茄串生长特性,提出基于实例分割的番茄串视觉定位与采摘姿态估算方法。首先基于YOLACT实例分割算法的实例特征标准化和掩膜评分机制,保证番茄串和果梗感兴趣区域(Region of interest,ROI)、掩膜质量和可靠性,实现果梗粗分割;通过果梗掩膜信息和ROI位置关系匹配可采摘果梗,基于细化算法、膨胀操作和果梗形态特征实现果梗精细分割;再通过果梗深度信息填补法与深度信息融合,精确定位采摘点坐标。然后利用果梗几何特征、八邻域端点检测算法识别果梗关键点预测果梗姿态,并根据果梗姿态确定适合采摘的末端执行器姿态,引导机械臂完成采摘。研究和大量现场试验结果表明,提出的方法在复杂采摘环境中具有较高的定位精度和稳定性,对4个品种的番茄串采摘点平均识别成功率为98.07%,图像分辨率为1280像素×720像素时算法处理速率达到21 f/s,采摘点图像坐标最大定位误差为3像素,深度误差±4 mm,成功定位采摘点后采摘成功率为98.15%。与现有的同类方法相比,采摘点图像坐标定位精度提高76.80个百分点,采摘成功率提高15.17个百分点,采摘效率提高31.18个百分点,满足非结构化种植环境中番茄串采摘需求。
文摘采摘点的识别与定位是智能采摘的关键技术,也是实现高效、适时、无损采摘的重要保证。针对复杂背景下番茄串采摘点识别定位问题,提出基于RGB-D信息融合和目标检测的番茄串采摘点识别定位方法。通过YOLOv4目标检测算法和番茄串与对应果梗的连通关系,快速识别番茄串和可采摘果梗的感兴趣区域(Region of Interest,ROI);融合RGB-D图像中的深度信息和颜色特征识别采摘点,通过深度分割算法、形态学操作、K-means聚类算法和细化算法提取果梗图像,得到采摘点的图像坐标;匹配果梗深度图和彩色图信息,得到采摘点在相机坐标系下的精确坐标;引导机器人完成采摘任务。研究和大量现场试验结果表明,该方法可在复杂近色背景下,实现番茄串采摘点识别定位,单帧图像平均识别时间为54 ms,采摘点识别成功率为93.83%,采摘点深度误差±3 mm,满足自动采摘实时性要求。
