利用二维数字图像估算种猪体重

为解决生产过程中种猪体重称量困难,难以对其生长进行实时监测的问题,提出利用二维数字图像对种猪体重进行估算的方法。利用数码摄像机获取种猪样本图像,采用域值分割法对图像进行分割,根据投影区域与参考系的比例关系估算得到种猪真实投影面积和体高,并建立了其与体重的回归方程。实验结果表明:用该方程估算得到的种猪体重与实际称量体重的平均相对误差为3.2%,精度较高,验证了本方法估算种猪体重的可行性。本方法可以避免传统称量方法可能导致的猪的应激反应给生产带来的损失,降低种猪生产成本,可用于种猪生长过程的监测和研究,为种猪的现代化饲养管理提供了一种有效的监测手段。

Determination of pig weight from 2D images

Aiming at the problem in estimating the weight of a pig by the traditional methods,we developed a new method to weight pigs by computer vision technique,in which a pig's weight was estimated by the projected areas and heights.Compared with a pig's real weight,the mean relative error of the estimated weight was 3.2%.The experimental results indicate that this hands-off method has a great significance for scientific management of pig production and does not require large labor and material resources.