| 基于计算机视觉的水产动物视觉特征测量研究综述 |
| |
| 引用本文： | 段延娥,李道亮,李振波,傅泽田.基于计算机视觉的水产动物视觉特征测量研究综述[J].农业工程学报,2015,31(15):1-11. |
| |
| 作者姓名： | 段延娥 李道亮 李振波 傅泽田 |
| |
| 作者单位： | 1. 中国农业大学信息与电气工程学院，北京 100083; 2. 北京农学院计算机与信息工程学院，北京 102206;,1. 中国农业大学信息与电气工程学院，北京 100083; 3. 中国农业大学北京市农业物联网工程技术研究中心，北京 100083;,1. 中国农业大学信息与电气工程学院，北京 100083; 3. 中国农业大学北京市农业物联网工程技术研究中心，北京 100083;,4. 中国农业大学工学院，北京 100083; |
| |
| 基金项目： | 公益性行业（农业）科研专项项目（201203017）；国家国际科技合作专项项目（2013DFA11320） |
| |
| 摘 要： | 作为水产养殖集成信息化管理的主要信息源,水产动物视觉属性信息的测量不仅是判定水产动物生长状况,调控水质环境的主要信息依据,也是对水产动物进行喂养、用药、捕获、选别和分级等操作的前提基础。近年来,计算机视觉技术作为一项快速、客观、无损的检测方法,已被逐渐用于水产动物视觉属性的测量中,许多研究学者开展了大量的研究工作。该文更新和总结了国内外近20多年来有代表性的相关研究和解决方案,在描述计算机视觉检测系统的概念和组成结构的基础上,围绕尺寸测量、形状分析、颜色识别和质量估计等方面详细分析了计算机视觉技术在水产动物(以鱼类为主)视觉属性测量方面的国内外研究现状,着重阐述总结了研究人员在水产动物视觉检测的图像采集、轮廓提取、特征标定与计算等方面的具体改进措施,并对基于计算机视觉测量的水产动物疾病诊断,识别分类等综合应用现状也进行了分析探讨,以评估计算机视觉技术在水产动物视觉质量检测领域的总体应用情况和现存的主要问题,同时给出了今后的研究趋势与发展方向。
|
| 关 键 词： | 水产养殖 计算机视觉 视觉 测量 图像处理 品质检测 |
| 收稿时间： | 6/7/2015 12:00:00 AM |
| 修稿时间： | 2015/7/20 0:00:00 |
Review on visual attributes measurement research of aquatic animals based on computer vision |
| |
| Institution: | 1. College of Information and Electrical Engineering, China Agricultural University, Beijing 100083, China; 2. College of Computer and Information Engineering, Beijing University of Agriculture, Beijing 102206, China;,1. College of Information and Electrical Engineering, China Agricultural University, Beijing 100083, China; 3. Beijing ERC for Internet of Things in Agriculture, China Agricultural University, Beijing 100083, China;,1. College of Information and Electrical Engineering, China Agricultural University, Beijing 100083, China; 3. Beijing ERC for Internet of Things in Agriculture, China Agricultural University, Beijing 100083, China; and 4.College of Engineering, China Agricultural University, Beijing 100083, China; |
| |
| Abstract: | Abstract: In aquaculture, visual attribute information of aquatic animals is the basis of determining growth condition, feed conversion, medication dosage, harvesting date and grading for aquaculture farmers and managers. For improving the quality of aquatic products, the automatic and non-destructive measurement of visual attributes is becoming more and more important in modern fishery. For decades, computer vision, as a non-destructive, rapid, economic, consistent, reliable and objective inspection tool based on image analysis and processing with a variety of applications, has been gradually used in visual quality detection of aquatic animals. Quite a number of researches have highlighted its potential application in aquaculture. Underwater or overwater video/image measurement systems based on image processing technologies have been used widely for automatically counting and measuring fish in aquaculture, fisheries and conservation management. However, the application of computer vision technologies in aquaculture is very challenging because the inspected objects are sensitive, easily stressed and free to move in an environment in which lighting, visibility and stability are generally not controllable, and the camera must be operated underwater or in a wet environment. This review updates and summarizes recent representative researches and industrial solutions proposed in order to evaluate the general trends of computer vision and image processing in the visible range applied for inspection of aquatic animals. On the basis of introducing the mode of operation and the components of a computer vision detection system, this paper presents a review of the overseas and domestic research status in visual attribute measurement of aquatic animals according to inspection tasks that are common to almost all visual attribute detection systems of aquatic animal: measurement of size and shape parameters, estimation of mass and quantification of color, etc. Specially, the techniques involve in computer vision detection system used for the improvements of data acquisition environment, accuracy of object detection and contour extraction, and the measuring results are analyzed in detail, including the consideration of image acquisition method and mode, the development of fish detection and feature points definition algorithm, as well as the study about feature computation method and mass prediction model. In addition, the comprehensive application of computer vision detection technology in aquatic animals is also discussed, including disease diagnosis, identification of species, detection of gender and age, as well as grading and sorting of fish. The objective of the review is to highlight the development of computer vision systems, image analysis and processing approaches in aquaculture and analyze the advantages and limitations of current computer vision detection systems which have made some progresses, but have not matured into a useful tool in aquaculture. Considering the overall trends, we propose some future research directions of the computer vision detection systems for aquatic animals, including the technology of image acquisition in natural underwater environment, complete process of fish detection and contour extraction, seamless integration of modules, as well as the technology of multi-information fusion. With the future development in these areas, computer vision detection technique may achieve higher accuracy and efficiency, and wider application in visual quality detection of aquatic animals. |
| |
| Keywords: | aquaculture computer vision vision measurements image processing quality detection |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
| 点击此处可从《农业工程学报》浏览原始摘要信息 |
| 点击此处可从《农业工程学报》下载免费的PDF全文 |
|
