基于牛体点云的牛体尺自动测量算法  

Automatic measurement algorithm of bovine body size based on bovine body point cloud

作　　者：陈禹州 石炜[1] 张显宇 张万锴 张国英 CHEN Yuzhou;SHI Wei;ZHANG Xianyu;ZHANG Wankai;ZHANG Guoying(College of Mechanical Engineering,Inner Mongolia University of Science and Technology,Baotou 014010,China;Inner Mongolia Zhimu Traceability Technology Development Co.,Ltd.,Baotou 014010,China;Inner Mongolia Zhongnong Hi-tech Technology Co.,Ltd.,Ordos 017010,China)

机构地区：[1]内蒙古科技大学机械工程学院,内蒙古包头014010 [2]内蒙古智牧溯源技术开发有限公司,内蒙古包头014010 [3]内蒙古中农高科科技有限公司,内蒙古鄂尔多斯017010

出　　处：《黑龙江畜牧兽医》2023年第8期48-54,136,共8页Heilongjiang Animal Science And veterinary Medicine

基　　金：内蒙古自治区科技重大专项“基于物联网的内蒙古现代草原畜牧业生产监控及产品安全溯源平台建设”。

摘　　要：为了高效、准确地利用牛体点云数据自动测量牛只体尺指标,试验在牛体点云数据基础上提出了一种基于点云处理与图像处理融合的牛只体高、体长、胸围自动测量算法,该算法将点云坐标原点修改成牛体质心,通过随机采样一致性(random sample consensus,RANSAC)算法分割牛体与地面,利用法向量主成分分析(pricipal component analysis,PCA)修改牛体点云数据坐标轴;使用图像形态学处理并进行曲线拟合得到牛体点云俯视、侧视图轮廓曲线,通过轮廓曲线特征确定测量点并测量牛只体长、体高;再根据测量点通过直通滤波得到牛体胸围截面点云,通过膨胀、闭运算进行曲线拟合得到牛体胸围数据;利用自动测量算法、交互式测量、人工测量对21头牛只进行体长、体高、胸围体尺测量,并进行比较。结果表明:与交互式测量相比,自动测量算法测量牛只体高、体长、胸围的正确率分别为100%、95.24%、85.71%,平均误差分别为0.47%、0.59%、0.81%,并提升了测量速度;与人工测量相比,自动测量算法测量牛只体高、体长、胸围的平均误差分别为3.34%、3.74%、4.73%。说明自动测量算法提高了利用牛体点云数据测量牛只体尺的效率。In order to efficiently and accurately use bovine body point cloud data to automatically measure bovine body size indexes,an algorithm for automatic measurement of body height,body length and chest circumference based on the fusion of point cloud processing and image processing was proposed.In this algorithm,the origin of point cloud coordinates was modified into the bovine body center,the bovine body and the ground were divided by random sample consensus(RANSAC),and then the axes of the bovine body point cloud were modified by normal vector principal component analysis(PCA).Finally,the contour curve of the top view and side view of the bovine body point cloud were obtained by morphological processing and curve fitting.And the measurement points were determined through the existing contour curve features to measure the body length and height of the bovine body.According to the measurement points,the cross section point cloud of bull body chest circumference was obtained through filtering,and the data of bovine body chest circumference was obtained by curve fitting through expansion and closing operation.Finally,the body length,body height and boby chest circumference of 21 cows were measured and compared by automatic measurement,interactive measurement and manual measurement.The results showed that the accuracy of measuring body height,body length and chest circumference of cattle by automatic measurement compared to interactive measurement was 100%,95.24%,and 85.71%,respectively,and the average error between the automatic measurement and interactive measurement was 0.47%,0.59%,and 0.81%,respectively,and measurement speed of automatic measurement compared to interactive measurement was improved.The results indicated that the algorithm further improved the efficiency of measuring cattle body size by bovine body point cloud data.

关 键 词：牛 体尺 自动检测 点云处理 随机采样一致性 主成分分析 图像形态学 

分 类 号：S823.81[农业科学—畜牧学] TP79[农业科学—畜牧兽医]