Accurate 3D shape recovery of live cattle with three depth cameras

Abstract

Abstract The animal carcass is one of important indicators of the development of young cattles, therefore it is essential to follow it up. Only the animals with sufficient body frame and with well-muscled top can be successfully fattened to high body mass. In this work we propose a prototype of non-intrusive scanning system for recovery of live cattle 3D shape with three depth cameras. To obtain the highest precision in measuring of cattle shape, we use calibrated cameras, curve fitting algorithms for solving the problem of missing data owing to partial occlusion, and algorithms for accurate fusion of point cloud data from three cameras. The measured animal 3D shape can be used, for instance, for automatic and precise estimation of body dimensions of live animals and for predicting the body weight of individual cattle as well as for daily monitoring production capacity of cattle.