基于机器视觉的单侧视猪腹式呼吸表达

Abdominal respiration expression of side-view single pig based on machine vision

为了监测猪的呼吸急促,引入机器视觉技术构建单侧视猪的腹式呼吸表达。根据猪腹式呼吸时脊腹部分呈起伏的特点,从视频帧的猪轮廓图像中提取形心,确定感兴趣的脊腹线段,然后在垂直方向提取脊线与腹线的截距,接着,构建脊腹线截距描述子(RACID)作为度量脊腹轮廓波动的指标,最后,根据帧序列中随时间变化的RACID,得到猪腹式呼吸表达。逐段检测猪视频腹式呼吸频次,实验结果显示机器法和人工法相关系数为98.23%。

To monitor the tachypnea of pigs, the machine-vision is introduced to construct a model on the pig＇s abdominal respiration. Firstly, according to the marked dilating and shrinking at ridge-abdomen of pigs, tworidge-abdomen profile segments are ensured based on the centroid of pig＇s contour. Secondly, the intercepts between both segments are obtained in vertical direction. Thirdly, Ridge-Abdomen Contour Intercept Descriptor （RACID） is created as an index of the fluctuating ones. Finally, the pig abdominal breathing expression is created according to the varying RACID over time. The frequency of the pig abdominal respiration was respectively obtained by the machine-vision method and the artificial method. Experiments showed that the correlation coefficient between the two methods is 98.23%.