基于模糊逻辑和支持向量机的奶牛发情监测系统设计

Design of dairy cow estrus monitoring system based on fuzzy logic and support vector machine

无接触式及时监测奶牛发情,发现最佳配种时期有助于奶牛繁殖,提高动物福利。在奶牛发情期其运动量增大,爬跨行为增多,利用几何特征能够直观地描述爬跨行为。针对明显的爬跨行为,根据最小外接矩形框计算多维几何特征,将其作为支持向量机行为识别的重要依据。当识别出爬跨行为后,依据含有爬跨行为的帧数计算爬跨时长,最终将产生爬跨行为的时间和爬跨时长作为模糊逻辑推理模型的输入,发情奶牛最佳配种时期作为其输出。实验结果表明,提出的方法可对实际的视频进行有效处理,对奶牛行为识别的平均准确率能达到97.5%,最佳配种时期与人工决策结果具有98.9%的匹配率,该发情监测系统能够实现管控一体化,为养殖人员提供可视化决策,有效提高奶牛繁殖性能。

The non-contact timely monitoring of the estrus of dairy cows can find out the best breeding period and help dairy cows reproduce and improve animal welfare.During the estrus of dairy cows,the amount of exercise and climbing behavior increase,and the geometric features can be used to intuitively describe the climbing behavior.Aiming at the obvious climbing behavior,the multi-dimensional geometric features are calculated according to the smallest external rectangular box,which are taken as an important basis for the behavior recognition by support vector machine(SVM).After the climbing behavior is identified,the climbing duration is calculated according to the number of frames containing the climbing behavior.The time when the climbing behavior occurs and the duration of the climbing behavior are used as the input of the fuzzy logic inference model,and the best mating period of the dairy cows in the estrus is used as the output.The experimental results show that the proposed method can effectively process the actual video,its recognition accuracy rate of the behavior of dairy cows can reach 97.5%,and its match rate between the optimal mating period and the manual decision result reaches 98.9%.The estrus monitoring system can realize the integration of management and control,provide visual decision-making for poultry feeders,and effectively improve the reproductive performance of dairy cows.