Abstract:

[Objective] According to the characteristics of vast activity area, insensitivity to the environment and delayed movement of free range cattle population, by referring the existing animal body size measurement technology, a non-contact body size measuring system on the basis of unmanned air vehicle （UAV） off-line 3D reconstruction method capable of collecting cattle point cloud data was develped to establish a wide coverage, precise, and high-efficient cattle body size measurement solution. [Method] Firstly, a quad-axis UAV was used to capture images, and some image processing techniques such as noise reduction with median filter and image three-channel histogram equalization were utilized to preprocess the image and to improve the reconstructed point cloud density and measurement accuracy. Secondly, the aerial triangulation measurement was conducted using the position and orientation system （POS） information from the image, and the 3D model was reconstructed after the initial coordinate data of point cloud was confirmed by adjustment computation. Finally, the point cloud of individual cattle was extracted by using random sample consensus （RANSAC） algorithm in point cloud library （PCL） and then interactive measurement was made. [Result] Images from 4 cattle herds and 1 individual cattle were collected, and available point cloud of body size data from a total of 21 cattle were reconstructed. In comparision to the convetional manual measurement method, the average relative errors in body height, body length and chest circumference measured by the developed UAV based system were 3.87%, 4.16% and 5.06%, respectively. [Conclusion] The UAV based system developed in this study can be used to measure body size of medium- and low-density free range cattle herds in Inner Mongolia, and the error is within the acceptable range.

Key words: body size, non-contact measurement, unmanned aerial vehicle, image processing, 3D reconstruction, interactive measurement