畜禽行为及生理信息的无损监测技术研究进展

Review on noninvasive monitoring technology of poultry behavior and physiological information

畜禽信息主要包括动物健康信息、行为信息、情绪信息。禽畜养殖中,准确高效监测畜禽信息有助于分析动物的生理、健康和福利状况,及时发现生病或异常个体,以减少经济损失和保障动物福利。目前畜禽养殖中主要依靠人工观察方式获取畜禽信息,主观性强且精度低;或者在饲养过程中采用一些将装置植入动物体内或对动物进行手术的监测手段,造成动物应激反应,有损动物福利。无损监测技术可以有效减少人力,降低观察者对动物的影响,减少监测过程中对动物造成的损伤与应激反应,提高动物福利。随着信息技术的进步,畜禽信息无损监测技术也在不断发展。该文阐述了畜禽养殖中传感器监测、图像监测及声音监测3种无损监测技术在获取畜禽信息方面的研究与应用现状,并分析3种无损监测技术的优劣。传感器监测技术发展较其他2种技术相对更加成熟,应用也更加广泛,可用来监测动物饮食、行为姿态等,但适合动物穿戴、可长期高效工作的传感器节点技术有待突破;图像监测技术利用前景提取、模式识别等方法对动物图像进行分析,可进行动物行为识别、质量估计等,对动物影响最小。但目前算法还不成熟,装置受环境干扰较大,因此应用有限;声音监测技术起步较晚,受限于环境噪声的影响,识别正确率较低,但在动物行为监测、疾病预警、情绪识别、饮食监测等方面均有较好的应用前景。该文还展望了畜禽信息无损监测技术未来精准、高效、智能、经济的发展趋势。

The behavioral information and body conditions of animals are significant in precision livestock farming. And they have a considerable relationship with animal's welfare and diseases. Therefore, perceiving animals' body and behavior information harmlessly is critical to livestock breeding. A research review of diseases detection, body conditions detection, individual identification, behavioral analysis, and so on with noninvasive monitoring technologies was presented focusing on some prevalent livestock, including pigs, cows, sheep and chicken. And a summary of the advantages and disadvantages of 3 noninvasive monitoring technologies, i.e. sensor monitoring, image monitoring and sound monitoring in all the aspects was presented in this review. Sensor monitoring has been applied in the monitoring of feeding and drinking behaviors of animals, the identification of location of free-ranging animals and daily behaviors monitoring. Various sensors such as temperature transmitter and acceleration transducer have been used for years, so sensor monitoring is more reliable compared with the other 2 technologies. However, it is hard to design stable and suitable sensors which can work for a long period of time under the bad conditions in animal husbandry. As for image monitoring, it has been applied in the estimation of weight and body contour of animals, behaviors monitoring and body temperature measurement. Images of animals are acquired by cameras and thermal infrared imager and then processed with different methods to mine information. Although image monitoring influences animals least, it is susceptible to lighting conditions sometimes. Algorithms need to be developed to improve accuracy of image identification and reduce environmental influence. Besides, sound monitoring in animal husbandry has been applied in diseases detection, emotional state recognition, daily behaviors monitoring and estimation of feed intake of free-ranging animals. Calls of animals can be easily obtained with microphones, while meanings and contents of which are essential to understand. Feature parameters and methods are fundamental to get animal's sound meaningfully. The combination of Mel Frequency Cepstrum Coefficent(MFCC) and Hidden Markov Model(HMM) is proved to have good performance. Sound monitoring technology shows good identification performance in laboratory, while it is not as good as what scholars think due to the noisy animal husbandry filled with people talk, noise of clanging doors and wind. Thus, there is a need to update algorithms to improve identification accuracy in animal husbandry. Those 3 monitoring technologies are harmless to animals during the process of monitoring, while some supervision methods now available worldwide require device implanting or operation to the livestock, which is hence detrimental for increasing welfare. Whereas for the noninvasive monitoring technology, it can effectively cut down the manpower consumption, reduce the damage and stress response during the monitoring, lower the influence on the animals caused by observer, and then enhance the animal welfare. Sensor monitoring, image monitoring and sound monitoring perform well in different ways. It is worth a try to combine 2 or 3 of them to realize better monitoring performance in animal husbandry. Many attempts of noninvasive monitoring have been made and many products have been applied in some western countries, while Chinese scholars attempted to study it just decades years ago. Considering this, Chinese scholars should learn from western scholars and develop advanced noninvasive monitoring equipment.