摘要
为了提高生猪头部在线检测的实时性,实现基于热红外视频的生猪耳根体表温度在线监测,提出一种基于最优步长的生猪头部快速检测方法。首先在生猪头部左右两侧的可能运动区域,设计5条动态扫描线,从通道入口开始扫描头部的运动状态。在生猪头部左右两侧可能的运动区域,沿水平方向分别设计生猪头部检测框的左边线和右边线。在生猪头部前侧可能的运动区域,沿垂直方向依次设置了生猪头部检测框的下框动态扫描边线、垂直区间动态扫描线和垂直区间动态扫描线的下限线;其次,当生猪进入通道时,将高温阈值分别与左边线和右边线的温度进行比较,计算左框和右框的动态扫描线是否需要水平平移,进而确定生猪头部检测框体左边线和右边线的位置;最后,将高温阈值与垂直区间动态扫描线的温度进行比较,计算出最优垂直移动步长,进而分别确定生猪头部检测框体上边线和下边线的位置,实现基于最优步长的头部快速检测。利用采集到的40头生猪视频数据,在Matlab及C#平台上进行了测试,并与骨架扫描策略、压缩感知、核相关滤波等方法进行对比分析。结果表明,本文方法检测平均帧速分别比骨架扫描策略、压缩感知方法提高了74.4%和54.1%,检测精度比压缩感知、核相关滤波分别提高了11.03、13.82个百分点,耳根温度平均误差为0.235℃。
To improve the on-line monitoring rate of pig ear skin surface temperature and realize the long-term monitoring of pig ear skin surface temperature,a fast detection method based on optimal step for pig head was proposed.Firstly,five dynamic detection lines were designed to scan at the entrance of the channel.Secondly,as soon as the pig entered the channel,the optimal vertical step size was calculated by using the high temperature threshold and two vertical dynamic detection lines,so as to determine the position of the frame on the left and right sides of the pig head box.Finally,the results of the comparison between the high temperature threshold value and the temperature of the dynamic scan line in the vertical interval were used to calculate the optimal vertical movement step,and then the positions of the upper and lower edge lines of the pig head detection box were determined respectively,so as to realize the fast detection of the head based on the optimal step.The video data of 40 pigs were collected and tested on Matlab and C#platforms.The results showed that the average frame rate of the proposed method was 74.4%and 54.1%higher than that of the skeleton scanning strategy and the compressed sensing method,respectively.Compared with compressed sensing and kernel correlation filtering,the detection accuracy was improved by 11.03 percentage points and 13.82 percentage points,respectively.The mean error of ear base skin surface temperature was 0.235℃.The research result can provide technical support for the integration of the automatic detection system of pig body surface temperature.
作者
马丽
张旭东
冯彦坤
李彦超
刘刚
MA Li;ZHANG Xudong;FENG Yankun;LI Yanchao;LIU Gang(Key Laboratory of Agricultural Information Acquisition Technology,Ministry of Agriculture and Rural Affairs,China Agricultural University,Beijing 100083,China;College of Information Science and Technology,Hebei Agricultural University,Baoding 071001,China)
出处
《农业机械学报》
EI
CAS
CSCD
北大核心
2021年第S01期291-296,共6页
Transactions of the Chinese Society for Agricultural Machinery
基金
国家重点研发计划项目(2016YFD0700204)
