Research On Cattle Scale Measurement Method Based On Three-dimensional Point Cloud

Under the premise of modern social development,due to the popularity of computers and the prevalence of the Internet,various fields of China’s economic growth have become inseparable from science and technology.Especially in the livestock industry,the traditional production mode no longer meets the requirements of economic development,and precise livestock and welfare livestock gradually become essential elements of the story.The traditional cattle body measurement technology no longer meets the requirements of modern precise welfare animal husbandry,which requires no harm to livestock,automatic extraction,and intelligent measurement,and to realize its provisions,the thesis proposes a method of automatic cattle body measurement based on the 3D point cloud.The research of this thesis contains three parts: pre-processing of cattle point cloud data,acquisition of critical points of the cattle point cloud,and design of automatic cattle body measurement.First of all,the cattle point cloud data pre-processing,this project collected the 3D point cloud data of the laboratory model cattle,three steps of cattle 3D point cloud data pre-processing,including point cloud outlier removal,point cloud ground segmentation,point cloud background removal,through the above three steps of pre-processing to remove the noise point cloud,only retain the test subject cattle point cloud.Secondly,we use deep learning technology combined with 3D point cloud processing to obtain key points of a bovine point cloud.By comparing two methods of obtaining a contour curve by overall point cloud projection and contour curve by point cloud slicing,we finally choose the method of securing contour curve by slicing bovine point cloud,analyze the change of local contour feature area of the bovine body,and obtain the spatial position of key points of each size by combining the change of curvature of contour curve The coordinates of the key points of each individual scale were obtained by blending the changes of curvature of the contour curve.Finally,the spatial position coordinates of the key points of the body scale associated with each body scale parameter were used to calculate the body scale parameters,and the error of bovine body slant length was 1.10%? the mistake of bovine body height was3.09%? the error of bovine body straight length was 0.99%? the error of bovine body depth was 5.97%? the mistake of bovine shoulder width was 9.03%? the error of bovine abdominal width was 8.15%.In this thesis,we combine cattle body measurement and a 3D point cloud,and design an automatic measurement method for cattle body measurement based on the processing of a 3D point cloud.The results show that the technique can automate cattle body measurement and obtain specific body parameter results,which is of great reference significance for herders to grasp cattle body size and its health condition,accurate welfare,and scale feeding in the future.The results show that the method can automate the measurement of body size and obtain specific body size parameters.