摘要
机器视觉技术已广泛应用到农业生产的诸多领域。综合国内外优秀研究成果,阐述了现阶段机器视觉在农业方面应用的主要形式,介绍了机器视觉在农作物精选与质量检测、植物生长信息监测、农田视觉导航等应用方向的研究成果,通过分析其创新性的图像处理算法、机器视觉系统的组成,提出了当前机器视觉农业应用仍存在可靠性差、成本高、智能化水平不高等问题。结合当前机器视觉在各种领域的研究和应用情况,对未来机器视觉在农业应用的发展方向进行展望,认为基于嵌入式处理模块和多技术融合的机器视觉系统将成为未来主要发展趋势,以卷积神经网络为代表的深度学习模型也将成为未来图像识别的核心技术,并将极大改善目前机器视觉在农业应用存在的诸多问题。
Machine vision technology has been widely applied to various fields of agricultural production. In this paper we summarize the outstanding research results at home and abroad and elaborate the main forms of machine vision application in agricuhure at this stage. We list the research results of machine vision in crop selection and quality inspection, plant growth information monitoring, field vision navigation, and other application directions. By analyzing its innovative image processing algorithms and the composition of machine vision systems, we show that the current application of machine vision in agriculture is still having problems such as poor reliability, high cost, and low level of intelligence. We analyze its subjective and objective reasons and put forward corresponding suggestions. Combining the current research and application of machine vision in various fields, we look forward to the future development direction of machine vision in agricultural applications. We believe that machine vision systems based on embedded processing modules and multi-technology integration will become the main development trend in the future. The deep learning model represented by convolution neural network will also become the core technology of future image recognition, which can greatly help solve the existing problems of machine vision in agricuhural applications.
作者
陈兵旗
吴召恒
李红业
王进
CHEN Bingqi;WU Zhaoheng;LI Hongye;WANG Jin(College of Engineering, China Agricultural University, Beijing 100083, China)
出处
《科技导报》
CAS
CSCD
北大核心
2018年第11期54-65,共12页
Science & Technology Review
基金
国家自然科学基金项目(31071329)
关键词
机器视觉
农业
无损检测
导航
machine vision
agriculture
nondestructive testing
navigation
作者简介
陈兵旗，教授，研究方向为图像处理，电子信箱：fbcbq@163．com
