高级检索+

基于Kinect相机的猪弯曲体尺测量算法研究

Research on the algorithm of curved body size measurement of pig based on Kinect camera

  • 摘要: 为解决猪体弯曲状体尺测量难、需借助标定尺还原真实体尺的问题,以长白猪为研究对象,基于Kinect相机获取猪体俯视和侧视深度图像,提出一种无需标定尺的猪体弯曲体尺测量方法。利用凸包络和最大外接轮廓,分别提取猪体头尾部、肩臀部和体高体尺测点,测算生猪体尺参数(体长、肩宽、臀宽、体高)。结果表明,在猪体正常弯曲范围内(0°~40°),肩宽测量精度最高,平均相对误差不高于1.92%,臀宽、体高和体长平均相对误差不高于2.00%、2.64%和2.93%;在猪体非弯曲状态下,肩宽、臀宽、体高和体长平均相对误差分别为1.29%、1.26%、1.92%和2.33%,测量精度均优于现有猪体体尺测量方法。综上,文章提出的方法可实现猪体自然状态下体尺测量,且精度较高,同时具有良好普适性和鲁棒性,满足生猪实际生长状况监测需要。

     

    Abstract: To solve the difficulty of measuring curved body size of pigs and restoring the real body size of pigs with calibrated ruler, landrace was taken as the research object, based on the top and side view depth images of pigs, this paper proposed a method for curved body size measurement of pigs without calibrated ruler. The method used convex envelope and maximum external contour to extract the body measurement points of head, tail, shoulder, hip, and body height, and then measured the body size parameters(body length, shoulder width, hip width, and body height) of the pig. The results showed that in the normal bending range(0°-40°), the measurement accuracy of shoulder width was the best,with less than 1.92% the average relative error, and the average relative error of hip width, body height and body length was less than 2.00%, 2.64% and 2.93%, respectively; in the non-bending state, the average relative error of shoulder width, hip width, body height and body length was 1.29%, 1.26%,1.92% and 2.33%, respectively, which was better than that of the existing methods. In conclusion, the method proposed in this paper can achieve the measurement of body size in the natural state of pigs with high accuracy, good universality and robustness, which can meet the needs of monitoring the actual growth condition of pigs.

     

/

返回文章
返回