玉米种子自动取样中籽粒分离与定向问题研究
文献类型：学位论文
| 作者 | 李广伟
|
| 学位类别 | 硕士 |
| 答辩日期 | 2016-05-25 |
| 授予单位 | 中国科学院沈阳自动化研究所 |
| 导师 | 谷侃锋 |
| 关键词 | 玉米育种 自动分离 定向 机器视觉 姿态识别 |
| 其他题名 | Research of Grain Separation and Orientation in the process of Maize Seed Automatic Sampling |
| 学位专业 | 机械制造及其自动化 |
| 中文摘要 | 农作物育种效率的提高是保障我国粮食安全的一个重要途径，育种样片自动制取装备研制成为了育种技术发展的一大障碍。本文针对育种用玉米种子切片自动制取装备研制过程中玉米籽粒单粒分离和自动定向问题进行了深入研究，设计了完整的解决方案，对关键部件进行了基于ANSYS的仿真分析以优化设计，搭建了实验平台并进行了验证实验。玉米籽粒分离和定向的准确度直接影响到切片切削位置和切片大小，是制约分析样品制取效率的关键环节，而玉米籽粒质量轻，体积小，形态、尺寸上一致性较差，常规分离和定向方法难以满足需求，本文采用机械装置与机器视觉相结合的方法，通过玉米籽粒单粒分离装置和玉米籽粒自动定向装置实现玉米籽粒分离与定向过程。玉米籽粒自动分离包括初步分离和精确分离两个过程，分离装置设计的核心内容是通过直线送料器和单粒分离器实现玉米籽粒精确、可控的单粒分离，单粒分离器安装在直线送料器料道末端，使用ANSYS仿真软件对料道结构和安装位置进行优化设计，实现玉米籽粒在料道入口附近分布集中而出口附近玉米籽粒分散的分离效果，以便单粒分离器有效分离，料道仿真优化结果与实验结果一致。使用一种计算机视觉与姿态调整装置相结合的定向方法，在图像采集器进行图像采集前进行一次机械对中，将玉米籽粒姿态限制在4种已知姿态以内，以弥补姿态识别中使用的神经网络分类有限性的缺陷。姿态调整完成后进行第二次机械对中，对玉米籽粒姿态进行修正，进一步减小了定向误差。视觉识别过程中使用MATLAB神经网络模式识别工具箱，通过训练神经网络将玉米籽粒不规则的形态特征与实际形态建立起一种模糊对应关系，识别过程中机械对中装置的参与起到了增强识别准确率，减小误差的作用，实现了机械装置与机器视觉的灵活交互。对497组样本进行训练，得到形态识别准确率97.8%，姿态识别准确率99.8%，识别速度1.3s/幅，能满足设备的设计要求，验证了姿态识别方法的准确性和姿态调整装置设计的合理性。 |
| 英文摘要 | The improvement of crop ? breeding efficiency is an important way to ensure the security of food in our ? country, automatic equipment technology used in sample preparation has become ? a major obstacle to the development of breeding technology. The problem of ? single kernel separation and automatic orientation of maize kernel use in the ? development of automatic production equipment of maize seeds was studied in ? this paper, a complete solution is designed, the simulation based on ANSYS of ? the key component was done, in order to optimize the design, test platform ? was set up and the experiment was carried out. The accuracy of separation ? and orientation of Maize Kernel directly affects the cutting position and ? size of the chip, it is the key link to analyze the efficiency of sample ? preparation, but the quality of corn grain is light, the volume is small, ? uniformity in shape and size is poor, conventional separation and orientation ? methods are difficult to meet the demand, the method of combining mechanical ? device with machine vision was used in this paper, separation and orientation ? of maize kernel by single realized through the corn kernel single grain ? separation plant and corn grain automatic orientation device. The automatic separation of ? maize kernel consists of two processes, which are preliminary separation and ? precise separation, the accurately and controllable of separation of maize ? kernel use linear feeder and single particle separator is the core content of ? the design of the separation device, single particle separator is installed ? in channel terminal of the linear feeder, using ANSYS simulation software to ? optimize the design and the installation position of the structure, to ? realize the separation distribution effect of maize kernel, that focus in the ? near entrance and dispersion near the exit of the channel, so that the single ? particle separator can works effectively, the simulation results of the ? material channel are in agreement with the experimental results. One of corn seed orientation ? method combined with computer vision and posture adjustment device is ? proposed in this paper, before the process of image acquisition, an alignment ? adjustment operation was done by mechanical adjusting device, the maize grain ? posture is limited to four kinds of known posture, to make up for the ? limitation of the neural network classification used in the gesture ? recognition. After the completion of the second alignment adjustment ? operation of the mechanical adjusting device, and modify the posture of corn ? grain, and the orientation error will be further reduced. Pattern recognition ? toolbox of MATLAB neural network was used in visual recognition, by training ? the neural network, a fuzzy relation could be established between the irregular ? morphological characteristics and the actual posture of the maize grain, in ? recognition process, the participation of the mechanical adjusting device ? plays a role in enhancing the recognition accuracy and reducing the error, ? realized the flexible interaction between mechanical device and machine ? vision. The 497 groups of samples are trained, the morphology recognition ? rate is 97.8%, the posture recognition rate is 99.8%, and the recognition ? rate is 1.3s per image, which can meet the design requirements of the device, ? and verifies the accuracy of the method and the rationality of the design of ? the posture adjusting device. |
| 语种 | 中文 |
| 产权排序 | 1 |
| 页码 | 67页 |
| 源URL | [http://ir.sia.cn/handle/173321/19667]  |
| 专题 | 沈阳自动化研究所_装备制造技术研究室 |
| 推荐引用方式 GB/T 7714 | 李广伟. 玉米种子自动取样中籽粒分离与定向问题研究[D]. 中国科学院沈阳自动化研究所. 2016. |
入库方式： OAI收割
来源：沈阳自动化研究所
浏览0
下载0
收藏0
其他版本
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。