| 摘要: |
| 【目的】设计一套机器视觉系统，用于实时测量各穴孔中幼苗叶片面积、判断是否适合移栽作业、确定适合移栽幼苗的抓取位置，为实现穴盘幼苗自动移栽作业奠定基础。【方法】用CCD数字摄像机采集番茄幼苗图像，转换成灰度图像，将幼苗与背景分割得到二值图像，去噪处理后，通过计算每个穴孔中幼苗叶片的面积来确定适合移栽的单元，并用形心法确定机械手抓取位置。【结果】采用1.8G－1.5R－1.8B灰度化因子、Otsu法分割幼苗与背景图像效果较好；采用单连通区域法统计幼苗叶片面积，经修正后相对误差小于1.0%，相对误差值平均下降了87.6%。【结论】设计的机器视觉系统具有较高的测量精度，能够满足穴盘幼苗自动移栽作业要求。 |
| 关键词: 机器视觉 自动移栽 图像分割 Otsu |
| DOI： |
| 分类号: |
| 基金项目:江苏省农机三项工程项目（NJ2008-32）；江苏省科技攻关计划重大项目（BE2006302） |
|
| Recognition and localization of plug seedling based on machine vision |
|
|
| Abstract: |
| 【Objective】This paper presents a machine vision system for transplanting plug seedling automatically.The system can be used for real time measuring the blade area of plug seedlings,sizing up whether it is suitable to transplant and determining the position of grabbing seedlings.【Method】The images of tomato seedlings captured by CCD digital cameras were changed into gray level images.Binary images were obtained by segmenting seedlings and backgrounds images.The denoising algorithm was used for the pre-process of seedling image.The cells suitable to transplant were determined after calculating the area of each seedling in cells.At last,centroid method was used to calculate the position of grabbing seedlings.【Result】It was effective to use the method of color characteristics value(1.8G-1.5R-1.8B)and Otsu algorithm to segment seedlings and image backgrounds.The relative error was less than 1.0% and it was reduced by 87.6% on average after correcting the leaves' area and using the simply connected region method.【Conclusion】The result shows that this machine vision system has high measurement precision,and can satisfy the requirement of automatic plug seedlings transplanting. |
| Key words: machine vision automatic transplanting image segmentation Otsu |
