【摘要】 在进行作物形态测量的研究中对叶片长度的测量要求较高的测量精度。而有关长度测量的算法较少,其中较为常用的是利用链码来进行曲线长度的测量算法,该算法要求对图像上的玉米叶子进行预处理之后,才能进行对叶片长度的测量,然而由于细化操作的误差和链码描述曲线的误差,无法满足对任意叶片长度的测量精度要求。基于以上需求,提出了一种利用抛物样条曲线和数值积分进行任意曲线长度测量的算法,该算法可以直接对图像上的玉米叶片进行长度测量,而测量之前无需对图像进行任何预处理,避免了细化操作和链码描述导致的误差,其测量精度满足了该项目的要求。经实验该算法的平均测量相对误差可以控制在1%以内。由于该算法理论上基于数学,所以具有较高的可靠性,可应用于对测量速度要求不高,而对测量精度要求较高,测量结果稳定性要求较好的其它项目之中。更多还原

【Abstract】 When carrying out the research of crop form survey,we are required the better survey precision by length measuring.There are fewer algorithm related to length measuring,the algorithm ″using chain code to measure the curve length″ is in common use,this algorithm needs to process the crop’s leaf on the image.After these operations,it can measure the leaf’s length.Since the error caused by thinning and chain code,it can not meet the survey precision.Considering these requirements,We put forward an algorithm of curve length using parabola and numerical integral.The algorithm can measure the crop’s leaf length on the image directly without any pretreatment.It avoids the error caused by thinning and chain code and its precision meets the survey precision required by the project.Through experiment,the algorithm’s average survey relative error can be controlled under one percent.Since the algorithm based on mathematics,it has better reliability.It can be used in other projects which require better survey precision and better stability and the projects do not need rapid survey speed.更多还原