| The development trend of large-scale pig breeding has increasingly higher requirements for the automation,informatization and intellectualization of breeding.The monitoring of individual pig information through information technology can not only enable the producers to timely grasp their physiological and health conditions,but also improve animal welfare and benefit of breeding enterprises to a certain extent.Body temperature is an objective reflection of the pigs in their various activities.It’s also a significant physiological indicator.Traditional temperature measurement of pigs is carried out manually.Generally,the rectal temperature of pigs is used to represent the body temperature.The artificial contact measurement is not only inefficient,but also easily causes the pig to stress response.and there is a risk of cross infection between humans and animals.Infrared thermography as an alternative non-contact body temperature measurement method has became a tool for identifying various physiology and pathology phenomena.In the field of infrared vision,the research is mainly based on the analysis of artificial infrared image of pigs,and no breakthroughs have been made in the automatic detection and analysis of body temperature.In view of the difficulty in obtaining the body temperature of pigs in the process of largescale pig breeding,this paper uses infrared technology,electronic information technology and computer vision technology to research the body temperature inspection of pigs,so as to provide equipment support and data basis for the later research on the pig health monitoring and warning model.It mainly includes:(1)Designed an inspection equipment of the pig body temperature based on infrared technology,It’s equipped with FLIR A310 as thermal infrared image collector,and integrated infrared mobile acquisition device and system control device to realize infrared image inspection and collection of pigs;(2)In order to automatically extract the body temperature of pigs,based on the domestic and international research,this paper puts forward the ear region of pigs as the sensitive area of body temperature,and explores an ear region detection and body temperature extraction algorithm for infrared images of pigs.The algorithm consists of three parts: infrared image preprocessing for pigs,ear region detection algorithm based on Faster R-CNN,and body temperature extraction method for infrared ear region of pigs.The temperature inspection equipment of pigs designed in this paper was deployed and experimented in Dongcheng Piggery,Xinxing County,Yunfu City,Guangdong Province.The pig thermal infrared images collected by the equipment were used for algorithm experiment and analysis.The experimental results show that: the body temperature inspection equipment of pigs constructed in this paper operates stably,the loss rate of image acquisition is 0.09%,and the periodic inspection accuracy of the ear region manually calculated is about 90%,which can effectively collect the pig thermal infrared images and provide equipment support for automatic temperature extraction in the following research.The 2000 thermal infrared images of pigs collected by the equipment were selected as the data set of Faster R-CNN based ear region detection algorithm in this paper,among which 1600 images were used for training and 400 images were used for testing,the detection accuracy of pigs ear region was 85.54%.Based on the method of body temperature extraction in the ear region,the maximum and average temperature of the ear region of the 3～5 pigs in the limit column were extracted and monitored.The scientific analysis of the temperature values was verified by analyzing the temperature curve.It provides a data foundation for the subsequent excavation analysis of pig health monitoring and early warning model,and has certain application value in the automatic monitoring of pig body temperature. |
PDF Full Text Request