水稻叶片几何参数图像视觉测量方法研究  

Study on the Geometry Parameter of Rice Leaf Measuring Method Using Image Vision Technology

作　　者：杨红云[1] 孙爱珍[1] 何火娇[1] 

机构地区：[1]江西农业大学软件学院/江西省高等学校农业信息技术重点实验室,南昌330045

出　　处：《湖北农业科学》2015年第17期4317-4320,共4页Hubei Agricultural Sciences

基　　金：国家自然科学基金项目(61363041);江西省教育厅青年科技项目(GJJ11082)

摘　　要：提出了一种基于图像视觉原理的水稻叶片几何形态参数测量计算方法,通过图像分割处理,统计参考物与目标叶片面积的像素数,计算叶片面积;通过求取参考物和目标叶片的最小外接矩形方法,计算叶片的长度和宽度值;利用基于4个方向模板运算的距离变换算法对茎叶夹角图像进行骨架信息提取,利用Hough变换对提取的骨架信息进行直线检测,并进行茎叶夹角计算。结果表明,该方法能够精确、快捷地对水稻叶片几何形态参数进行测量计算,与手工测量比较,叶片面积计算误差小于5%,长宽误差小于0.67%,茎叶夹角误差小于2%,能满足农学研究的要求。A measurement method of the rice leaf geometry parameter based on visual image theory was presented. The leaf area was calculated by image segmentation processing and statistical pixels of the reference objects and the target leaf area;the length and width parameters of rice leaf were calculated by finding minimum enclosing rectangle method of the reference objects and the target leaf; the skeleton information was extracted from a image of the angle between rice stem and leaf by the distance transform algorithm based on the four directions template operations, the straight lines were detected from the skeleton information by the Hough transform, and the angle between rice stem and leaf was calculated by calculating the an-gle between the straight lines. The results showed that the method can accurately and fast measure the rice leaf geometry pa-rameter, compared with manual measurement, leaf area calculation error was less than 5%, length and width error is less than 0.67%, leaf-stem angle error is less than 2%, could meet the requirements of agricultural research.

关 键 词：水稻叶片 几何参数 最小外接矩形 视觉测量 骨架提取 直线检测 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]