油菜角果数量及关键表型参数的自动化检测方法研究

为有效替代人工方式考种油菜、观测角果,研究了一种用于测量批量角果的数量和关键表型参数的自动化检测方法。设计了角果散铺和图像采集装置,利用拨动加振动的方式将堆积角果均匀散开并拍摄视频;使用二维码作为标记块,以有效地提取关键帧并拼接为包含全部角果的整幅图像;提出了基于凹点提取与匹配的图像分割方法,分割各种形态下的重叠角果,准确率达到98%以上。在关键表型参数测量中,利用了最大类间方差法以判断角果的正置或侧置姿态,以此估计角果横切面的近似椭圆长短轴,再计算角果的长度、表面积和体积。实验结果表明该方法具有很好的检测精度,对不同品种的油菜适应性较好,长度、表面积和体积的估计误差分别不大于2.9%、4.8%和5.0%。该方法可以有效替代人工方式的油菜考种,为相关农业科研领域提供基础数据。

Automated detection research for number and key phenotypic parameters of rapeseed silique

Silique observation and measurement are essential for rapeseed breeding.In this paper,an automat ic detection method was proposed to replace the traditional manual method.A device was designed to acquire video of scattered siliques,based on pulling and vibrating stacked siliques.Silique videos were extracted to frames,and then,by using QR code as marker blocks in image,the key frames containing all siliques were effectively extracted and spliced into individual intact.Crossed siliques in frame caused an error in measurement.A cutting method of crossed siliques image based on concave point extraction and matching was proposed.By this method,all kinds of crossed siliques could be identified with accuracy rate of 98.0%.In the measurement of phenotypic parameters,a core diameter Otsu method was proposed to judge the posture of silique,by which elliptic long and short axis of the cross section of the silique was estimated,and then length,surface area and volume of the silique were calculated.Results demonstrated good accuracy and adaptability to different varieties of rape by this method.Estimation error of length,surface area and volume were less than 2.9%,4.8%and 5.0%respectively.Thus the method could be an effective replacement of artificial way and provide key basic data for rapeseed and agricultural research fields.