A multi-posture adaptive method for measuring goat bodies dimensions using 3D point clouds in real-world applications

Body dimensions are recognised as critical indicators in the processes of fattening and breeding goats. However, manual measurement methods are both time-consuming and labour-intensive, often inducing significant stress in the animals and compromising their healthy growth. To address these concerns, a non-contact, high-throughput body measurement system for goats has been developed utilising multi-view three-dimensional point clouds. It employs an arched-channel device equipped with three depth cameras to collect three-dimensional point clouds. Firstly, a Multi-View Filter Reconstruction Processing Pipeline is proposed to filter, down-sample, segment, register, extract and reconstruct the surface of the target point cloud. Secondly, a Directional Positioning Continuous Segmentation Algorithm is introduced to identify key regions of the goat's body for precise segmentation. Finally, body dimensions are calculated from the identified key regions. The average error rates of various measurements, including body height, body oblique length, chest girth, chest depth, chest width, abdominal girth, abdominal depth, and abdominal width are 1.30 %, 2.73 %, 1.89 %, 2.43 %, 3.37 %, 2.11 %, 2.94 %, and 2.59 %, respectively. This algorithm provides a scientific and accurate way to measure the key body dimensions of goats. This algorithm demonstrates generalisability and robustness, being unaffected by environmental conditions or the diverse postures of the goats. This automated method for measuring goat body dimensions offers an efficient means of collecting phenotypic data during goat fattening, providing a convenient approach to precise and high-throughput body dimension assessments in goat breeding.