基于android的水稻叶片特征测量系统,基于Android的水稻叶片特征参数测量系统

摘要

【目的】提供基于Android的水稻叶片特征参数测量系统,为农学研究提供精准数据。【方法】系统设计主要分为图像获取、图像预处理和特征参数计算3部分:采用智能手机相机获取水稻叶片图像;通过灰度化、二值化和轮廓提取等操作对图像进行预处理;根据算法计算得到水稻叶片特征参数。【结果】精确度测试中,建立系统获取颜色分量(R、G、B)均值与各分量实际值的线性拟合模型,其决定系数分别为0.9803、0.9774和0.9805,均方根误差分别为1.086、1.413和0.8383;叶片几何参数中,系统获取水稻叶片长、宽、面积及周长与水稻叶片实际值的均方根误差分别为0.6469cm、0.05022cm、0.5329cm^2和2.412cm。耗时测试中,采用多部不同配置手机对同一叶片进行测试,图像预处理及参数计算总时间均在2s以内。可见,基于Android的水稻叶片特征参数测量系统能快速、准确、便捷地获取水稻叶片长、宽、面积、周长及颜色等特征数据,在一定程度上实现户外实地测量,满足农学研究的需要。【建议】对系统进行进一步优化,降低对外部辅助物的依赖,真正实现操作简便;增加氮素营养水平诊断算法实现氮素营养水平分类,以快速精准判断水稻生长营养状况,实现精确施氮。

【Objective】An android-based measurement system for characteristic parameters of rice leaves was created in order to provide accurate data for agricultural research.【Method】The system design was mainly divided into three parts:image acquisition,image pre-processing and character parameter calculation.A smart phone camera was used to capture images of rice leaf.The image was pre-processed by graying,binarization and contour extraction.The parameters of rice leaves were calculated according to the algorithm.【Result】In the accuracy test,The linear fitting models of color components(R,G,B)averages between predicted data(obtained by system)and actual data were established respectively and their decision coefficients were 0.9803,0.9774 and 0.9805,the root mean square errors were 1.086,1.413 and 0.8383.For geometry parameters(leaf length,leaf width,leaf area,leaf perimeter)of leaves,the root mean square errors between predicted data(obtained by system)and actual data were 0.6469 cm,0.05022 cm,0.5329 cm^2 and 2.412 cm,respectively.In the time-consuming test,the same blade was tested with multiple different configurations of mobile phones,and the total time of image pre-processing and parameter calculation was all within 2 s.The experiment results showed that the android-based system for measuring the characteristic parameters of rice leaves could quickly,accurately and conveniently obtain the characteristic parameters such as leaf length,leaf width,leaf area,leaf perimeter and color components.It could realize field measurement to certain extend,and met the demand of agricultural research.【Suggestion 】The following suggestions were proposed:further optimizing system to reduce the dependence on external auxiliaries to realize simple operation;adding nitrogen nutrition diagnosis algorithm to realize the classification of nitrogen nutrition level,so as to quickly and accurately judge the growth nutrition status of rice and accurately apply nitrogen.