沈明霞,李嘉位,陆明洲,刘龙申,孙玉文,李 泊.基于动态多特征变量的黄羽肉鸡跛行状态定量评价方法[J].农业机械学报,2018,49(9):35-44.
SHEN Mingxia,LI Jiawei,LU Mingzhou,LIU Longshen,SUN Yuwen,LI Bo.Evaluation Method of Limping Status of Broilers Based on Dynamic Multi-feature Variables[J].Transactions of the Chinese Society for Agricultural Machinery,2018,49(9):35-44.
摘要点击次数: 229
全文下载次数: 105
基于动态多特征变量的黄羽肉鸡跛行状态定量评价方法   [下载全文]
Evaluation Method of Limping Status of Broilers Based on Dynamic Multi-feature Variables   [Download Pdf][in English]
投稿时间:2018-03-28  
DOI:10.6041/j.issn.1000-1298.2018.09.004
中文关键词:  肉鸡跛行  步态评分  图像处理  步幅差  异常指数  决策树
基金项目:国家重点研发计划项目(2017YFD0701602-2)、国家自然科学基金青年基金项目(61503187)和常州市科技支撑计划项目(CE20172005)
作者单位
沈明霞 南京农业大学
江苏省智能化农业装备重点实验室 
李嘉位 南京农业大学
江苏省智能化农业装备重点实验室 
陆明洲 南京农业大学
江苏省智能化农业装备重点实验室 
刘龙申 南京农业大学
江苏省智能化农业装备重点实验室 
孙玉文 南京农业大学
江苏省智能化农业装备重点实验室 
李 泊 南京农业大学
江苏省智能化农业装备重点实验室 
中文摘要:肉鸡步态是肉鸡健康状态的重要表征,为实现黄羽肉鸡跛行步态的损自动化快速分类识别,提出了一种基于多特征变量的肉鸡跛行定量评价方法。该方法从步态视频中提取肉鸡的速度、步幅、步幅差、步频、投影面积参数,拟合得出速度、步幅与投影面积具有相关性(决定系数分别为0.8051、0.7935),据此定义肉鸡动态理想参数与异常指数,基于C45决策树模型,以速度异常指数、步幅异常指数、步幅差异常指数为分类特征,根据鸟类步态评分标准将肉鸡分为GS0~GS4五类,实现对肉鸡跛行状态进行预警和判别。实验结果表明:该模型针对GS0~GS4分类准确率依次为:66%、71%、74%、98%、95%,整体准确率为78%。该模型可作为早期跛行的检测工具,为养殖自动化的实现和动物福利产业的升级提供支持。
SHEN Mingxia  LI Jiawei  LU Mingzhou  LIU Longshen  SUN Yuwen  LI Bo
Nanjing Agricultural University;Jiangsu Provincial Key Laboratory of Intelligent Agricultural Equipment,Nanjing Agricultural University;Jiangsu Provincial Key Laboratory of Intelligent Agricultural Equipment,Nanjing Agricultural University;Jiangsu Provincial Key Laboratory of Intelligent Agricultural Equipment,Nanjing Agricultural University;Jiangsu Provincial Key Laboratory of Intelligent Agricultural Equipment,Nanjing Agricultural University;Jiangsu Provincial Key Laboratory of Intelligent Agricultural Equipment and Nanjing Agricultural University;Jiangsu Provincial Key Laboratory of Intelligent Agricultural Equipment
Key Words:broiler limping  gait score  image processing  step difference  anomaly index  decision tree
Abstract:The bird gait score (GS) is an important tool for evaluating the gait status of broiler. GS0~GS5 corresponds to the broilers whose limping level varies from low to high. The level of limping is used as an important indicator to measure the health of broilers. At present, traditional methods for gait assessment of broiler are mostly completed by visual inspection. The process is time consuming with low standardization. The dynamic feature variables extracted from video were used to evaluate the gait status of broilers based on decision tree, and a fast, stable and non contact broiler gait evaluation method was explored. The experiment was conducted at Quanjiao Broiler Breeding Center of Wenzhou Group, from December 2017 to January 2018. A total of 260 broilers (GS0~GS4) were selected. Each broiler was subjected to twice walking experiments. The experiment was conducted in a special broiler walkway. Two cameras were placed on the opposite side of the walkway and at the top of walkway, and videos were collected horizontally and vertically. Each frame of the video underwent image reorganization, filtering for pretreatment in HSV space. The broiler projection area was calculated by the least squares ellipse fitting based on vertical image, and the dynamic parameters such as the walking speed, stride length, stride difference value, and walking steps of the broilers were calculated based on horizontal image. Based on the study of the dynamic parameters of GS0 broilers, the linear fitting relationship between walking speed, stride length and projection area of broilers was obtained by the least square method, the coefficient of certainty was 0.8051 and 0.7935, respectively. According to the fitting results, based on the different top projection areas of the broiler, the ideal stride and ideal speed of the broiler were proposed. Then, according to the difference between the actual value and the ideal value of the parameters such as stride and speed, the abnormal index of dynamic parameters in broiler walking was defined. Taking the anomaly index, including speed, stride and step difference as training attributes, the C45 decision tree model was optimized for learning and post pruning. Totally 520 data was verified by a 10 fold crossover method to obtain the classification result. The accuracy of GS0~GS4 classification was 66%, 71%, 74%, 98% and 95%, and the overall accuracy was 78%. The above results showed that based on the dynamic multi feature variables extracted from video and decision tree model, the quantitative evaluation of limping state of broilers can be achieved. The research result provided a method for assessing the degree of non contact broilers with high accuracy. The method can be used as an early detection tool for identification and early warning for broilers limping, which provided support for the realization of farming automation and animal welfare industry upgrading, which had certain practical value.

Transactions of the Chinese Society for Agriculture Machinery (CSAM), in charged of China Association for Science and Technology (CAST), sponsored by CSAM and Chinese Academy of Agricultural Mechanization Science(CAAMS), started publication in 1957. It is the earliest interdisciplinary journal in Chinese which combines agricultural and engineering. It always closely grasps the development direction of agriculture engineering disciplines and the published papers represent the highest academic level of agriculture engineering in China. Currently, nearly 8,000 papers have been already published. There are around 3,000 papers contributed to the journal each year, but only around 600 of them will be accepted. Transactions of CSAM focuses on a wide range of agricultural machinery, irrigation, electronics, robotics, agro-products engineering, biological energy, agricultural structures and environment and more. Subjects in Transactions of the CSAM have been embodied by many internationally well-known index systems, such as: EI Compendex, CA, CSA, etc.

   下载PDF阅读器