基于计算机视觉的油菜叶面积计算方法研究

介绍了利用计算机视觉技术测量油菜叶面积的原理和方法,并以纸重法为标准对比分析了传统直尺法与此方法在油菜三种类型叶片叶面积测量中的优劣性。结果表明,计算机视觉法在精度上明显高于直尺法,尤其在无柄叶测量中此方法的测量平均相对误差为3.09%,低于直尺法的10.49%,进一步根据标准测量值对直尺法测量无柄叶时的系数进行校正,发现系数选择0.75更为合适。

Research on the Area Measurement of Rapeseed Leaves Based on the Computer Vision Technology

The principles and methodologies of brassica napus leaf area measurement by using computer vision technology were introduced in this paper.In the area measurement of the three types of rapeseed leaf,with the paper weight method as a standard,the accuracy of ruler and computer vision method was compared.The results showed that the computer vision method was superior to the ruler method in accuracy,especially for the no petiole leaf,with the average relative error of computer vision method at 3.09% which was significantly lower than ruler method,at 10.49%.According to the measured value of standard method,the corrected coefficient of ruler method for no petiole leaf was found that 0.75 is more appropriate.