【作者】 刘继承；

【导师】 姬长英；

【作者基本信息】 南京农业大学 ， 农业机械化工程， 2007， 硕士

【摘要】 依据水稻的生长状态信息对其生长因素进行控制，对于最大限度地节约能源并使水稻高产具有重要意义。针对传统的人工观察法效率低下且观察结果主观性强；遥感监测、远程监测等方法受观测范围所限且不适合单个农场实时快速、小面积监测等不足，本文提出利用图像处理技术对所拍摄到的水稻图像进行分析处理，并据此获取水稻的生长状态，进而实现对水稻长势进行监测。本文通过在南京浦口农场水稻田进行实地实验，利用数码相机获取水稻图像并建立了相应的图像处理方法。论证了依据单株水稻图像形态参数的提取来判断水稻长势状况的可行性，并据此法分别探讨研究了几个主要生育期群体水稻的叶面积及叶面积指数与实测的叶面积及叶面积指数间的相互关系模型。主要的研究内容如下：针对单株水稻图像叶尖部位不易分割出的特点，本文提出了采用局部灰度线性变换来对水稻图像实行增强处理；单株水稻图像细化过程中出现的毛刺现象在文中也采用了较好的处理方法进行去除；此外，比较了中值滤波法的几种滤波窗口对水稻群体图像的去噪效果，最终选用十字型中值滤波法去除图像采集过程中受到的各种噪声干扰；提出用颜色特征法对水稻群体图像进行叶片图像和背景的分割。株高、叶尖距以及叶基角是三个表征水稻长势的主导因素，而这些参数只通过图像处理可以比较容易、精确的获取，所以采用这三个参数作为图像处理技术对水稻长势监测的重要指标。而对于叶尖距，无论是运用手工测量方法还是图像处理方法对其测量，均不能用单个叶尖距的数值大小进行水稻长势状况优劣的判断，需要统计其平均叶尖距，并将其作为判断水稻长势优劣的依据。由于叶面积指数是作物长势诊断的主要参数，本文分析了9746水稻在不同生长期的叶面积指数的变化规律，并建立了图像叶面积指数与实际叶面积间的叶面积指数模型。通过检验后发现由模型得出的长势监测指标叶面积指数和真实叶面积指数的差异性不显著，故本文认为可以通过运用叶面积指数模型来获取植株的生长信息。综上所述，本文认为应用图像处理技术来获取水稻植株长势特征的方法是可行的。更多还原

【Abstract】 It is of great significance to control the growth factors according to the rice growth conditions information for saving energy and making high-yielding. The traditional manual observation is inefficient and the observed result can be easily effected by subjective factor. Remote sensing monitoring and remote monitoring are limited by the monitoring scope and cannot be used for real-time monitoring and small regions monitoring on individual farms. So this paper proposes method of rice growth monitoring based on image processing.The rice images were got by digital camera and the corresponding image processing methods were proposed from the experiments in Nanjing Agricultural University farm in this paper. The feasibility of judging the rice growth conditions based on single rice plant morphological parameters of the images was proved. The relationship between the leaf area and leaf area index in several growth periods and the relationship between the leaf area and leaf area index obtained by measurements was also studied. The main contents are as follows:This paper proposes the using of local gray linear transform for the rice image enhancement since the leaf tip of rice plants images were difficult to be segmented; A better method was used to eliminate the burr of the rice images appeared in thinning images; In addition, by the comparison of effectiveness of denoising of rice image for several median filtering filter window using median filter method, the cross-median filter was chosen to denoise during the image processing. It is proposed that the using color characteristics.The plant height and leaf basic angle are taken as the standard parameters to evaluate the rice growth conditions because that the plant height, leaf tip distance and leaf basic angle are the main factors to describe rice growth conditions. And rice growth status are also estimated by the leaf tip distance average datum.The leaf area index is the main factor to describe rice growth conditions. It is found that it’s feasible to obtain the rice growth condition information from leaf area index in several growth periods. There was little difference between the growth parameters obtained from the leaf area index model and the verity value. So, it can conclude that it’s feasible to obtain rice growth characteristics using image processing.更多还原