【作者】 许彩虹；

【导师】 张树君；

【作者基本信息】 北京交通大学 ， 电子与通信工程， 2012， 硕士

【摘要】 叶片是植物进行光合作用和蒸腾作用的重要器官,叶片发育情况以及面积大小对绿色植物的生长、抗逆性以及产量形成很大影响,同时也反映各个生长阶段是否需要对其施肥、浇水等。因此,正确快速测量叶片面积对农业生产具有重要的意义。然而,传统叶片面积测量方法中,大部分是对单叶片的测量,这种测量具有一定局限性,当需要对叶片进行无损测量时,也即不能破坏植物生产的连续性时,叶片之间可能具有重叠的部分,这就迫切需要设计出对重叠叶片面积测量的算法,以此来满足无损叶片面积测量的需求。本文就是针对传统叶片面积测量中存在的问题,同时在参考大量相关文献的基础上,提出了一种便携式重叠叶片面积测量仪的设计与实现的方案。该设计采用STM32作为硬件平台,选取OV7660为图像采集设备,对待测具有部分重叠叶片进行图像采集,对采集到的具有复杂背景的图像经过基于Mean-shift的图像边缘检测,再估计出叶片面积,最后把经过处理后的叶片图像和计算出的叶片面积显示在显示屏ILI9320上面。本文主要工作及其创新性主要体现在以下三个方面：1.在复杂背景下,先对采集到的图像进行Mean-shift平滑去噪处理,再根据目标叶片的颜色特征,利用超绿色灰度化对图像进行灰度化,提取出目标叶片,然后经提升小波变换增强弱边缘,最后再利用Canny边缘检测和二值化的处理,这样就可以从复杂背景中提取出目标叶片,同时也对具有部分重叠叶片进行分割。2.利用传统的参考物法计算出单个叶片的面积并存储,再根据同一种植物叶片在形状上符合fractal规律,可以根据待测重叠叶片中每一个叶片与计算出的单个叶片之间的比例关系,估计出重叠叶片中每一个叶片的面积,即可求出重叠叶片面积。3.把设计的重叠叶片面积计算的算法在PC机和STM32硬件平台上进行调试,实验结果证明了该算法的可行性和准确性。更多还原

【Abstract】 The leaves of plant are vital organs for photosynthesis and transpiration. To a certain extent, the growth index of the leaf playing an important role, the size of the leaf area hasn’t only an important role in the measurement of crop cultivation and breeding practices, but also reflected fertilization and watering or not in the various growth stages. Therefore, the correct and rapid measurement of leaf area is of great significance to agricultural production. However, the traditional leaf area measurements confined to the measurement of the monolithic leaves. When we need to measurement the non-destructive leaves, which may has overlapping, the traditional method has certain limitations. There is an urgent need to design the overlapping leaf area measurement algorithms, in order to meet the needs of non-destructive measurement of leaf area.In this thesis, I proposed a machine vision-based embedded measurement of overlapping leaf area, which based on the problems that exist in the traditional leaf area measurement and referred to the literature. The design uses the STM32as hardware platform, which acquires leaf image by OV7660. The processed image can be put into the STM32processor, displayed in the display module ILI9320.In this thesis, the work of its innovative results mainly reflected in the flowing three aspects:1. In the situation of complex background, using ultra-green gray can extract the target leaf, the after the lifting wavelet transform enhanced weak edge, and finally use the Canny edge detection and binaryzation treatment. Therefore, it can reach extracted from a complex background of the target leaves.2. We can calculate the area of single leaves and stored it. According to the same kind of plant leaves in the shape of the fractal law, we can estimate the each leaf area of the overlapping leaves.3. Through the calculated on the PC and the STM32hardware platform debugging, The experimental results demonstrate the feasibility and accuracy of the algorithm.更多还原