基于双目立体视觉的苗期玉米株形测

将田间正常生长的待测玉米植株带土移至测定台上,标定双目立体视觉系统,提取、分割叶片图像,以Douglas-Peucker多边形法逼近叶片边缘,去除两幅对应图像中没有匹配关系的多边形顶点,结合投影矩阵计算出叶片边缘点的三维坐标.分别投影叶片边缘点到植株平面和植株水平平面,对投影的离散点分段二次拟合、Cardinal样条插值,得到代表叶片形状的曲线,计算出叶长、叶片着生高度、茎叶夹角、叶片方位角等株形指标.测量实验表明,本方法快速、准确、自动化程度高,能够满足苗期玉米株形测量的要求.

of Maize Seedling Morphological Traits Based on Binocular Stereovision

Field-growing maize seedling was transplanted onto measuring board with soil reserved.Images of plant were captured with calibrated binocular stereovision system.The binary images were extracted and segmented into leaves and stems according to phyllotaxis.An algorithm of Douglas-Peucker polygonal approximation of leaf contour was applied to find matching points of two corresponding images.Projecting 3-D points of maize seedling onto plant plane and plant horizontal plane, fitting and interpolating the proje...