基于机器视觉黄瓜果实自动分级方法

摘要

To achieve the realization of grading cucumber fruit fast and accurately , we build an automation sub-base platform with the help of a camera as video acquisition module , DSP core processor as the main control module , as well as other auxiliary units like weight control and motor transport part .With reference to the national standard NY/T1587-2008 , we adopt the image processing method to extract and compute the melon length , pedicel length , diameter differ-ence , arcuate height of our cucumber fruit images , thus picking up 3 varieties 240 cucumber fruit as our test sample which contains 'Changchun Mici','Longzahuang VII'and ‘Luqiu I’.After, extract 20 samples of every variety as the image extracting data analysis and take the other 60 samples into the automation sub-base platform testing .Based on the test result , it shows: the average grade precision of the platform is 96 .7% while it can process 35 fruits per minute . Compared with the traditional manual classification , our platform has a characteristic of fast-processing , NDT, accurate and objective , which provides an important basis on applying machine vision technology into the automated classification of elliptical elongated fruit .%为实现黄瓜果实快速准确分级，以摄像头为视频采集模块、DSP 核心处理器为主控制模块、机械手为执行模块，并借助质量控制、电机传送等辅助单元，构建了自动化分级平台。参照国家标准 NY／T1587－2008，利用图像处理方法对黄瓜果实图像的瓜长、把长、横径差、弓形高度进行了提取和计算。选取长春密刺、龙杂黄七号、露秋一号3个品种240根黄瓜果实作为试验样本，抽取每个品种的20个样本作为图像提取数据分析，其余60个样本作为自动分级平台测试。测试结果显示：该平台的平均分级精度为96．7％，每分钟约检测35根果实，相较人工分级具有快速、无损、准确、客观的特点，为机器视觉技术应用于椭长形果实自动化分级提供了重要依据。