A posture-based measurement adjustment method for improving the accuracy of beef cattle body size measurement based on point cloud data

Abstract

Beef cattle body size is an important phenotypic trait for breeding and production. The non-contact automatic calculation of beef cattle body sizes via point cloud data exhibits several disadvantages, such as the inaccurate positioning of measurement points and the sensitivity to postures, heavily restricting commercial applications. To address these issues, an automatic method for key region segmentation, body size calculation, and data calibration to accurately measure the body size parameters of beef cattle in different postures is proposed. Firstly, the body sizes were obtained based on the point cloud data. Then the twelve micro-posture feature parameters were defined and extracted from four categories corresponding to the head, back, torso, and leg regions. Based on those parameters, a Posture-based Measurement Adjustment model was established to reduce the influence of postures on calculation results based on the correlation between measurement errors and micro-posture features. The experimental results show that, after the adjustment model calibration, the average errors of body oblique length, wither height, hip height, chest girth, and abdominal girth were 1.14%, 1.84%, 3.47%, 1.56%, and 2.36%, respectively, whereas the corresponding maximum errors were 4.36%, 8.10%, 11.10%, 7.29%, and 9.26%. The measurement errors of body sizes can be significantly minimised by the adjustment model, which can improve the robustness of different postures. The adjustment model improves the accuracy of the non-contact body size measurement, which could provide guidance and inspirations for future research.