摘要
[目的]本文旨在研究如何及时、准确获取生猪的活动信息。[方法]基于遗传算法和模拟退火算法对OTSU算法的阈值进行计算,并通过使用灰度直方图减少遍历迭代次数。再利用python广播机制,进一步提升了矩阵运算速率。[结果]使用改进的遗传算法和模拟退火算法对OTSU算法的阈值计算进行优化后,计算时间降低5%以上,误差不超过1%。[结论]优化后的算法为实时、准确的提取生猪运动、睡眠等活动状态提供了一定的借鉴和参考。
[Objectives]The study was aimed to accurately and timely obtain the information of the activities of live pigs.[Methods]Based on genetic algorithm and simulated annealing algorithm,the threshold of OTSU algorithm was calculated,and the number of iterations was reduced using gray histogram.Then the broadcasting mechanism in Python was applied to improve the rate of matrix operation.[Results]The experimental results showed that the threshold calculation time in OTSU algorithm was decreased more than 5% with the error less than 1% with the help of improved genetic algorithm based on simulated annealing algorithm and gray histogram and broadcasting mechanism.[Conclusion]The optimized algorithm can provide a reference for real-time and accurate extraction of pigs' activity status such as their movement and sleep.
引文
[1]沈富林,陆雪林,许栋,等.智能物联助推畜牧产业升级和种养联动[J].中国畜牧业,2016(1):32-35.
[2]罗土玉,边峰,钟日开,等.幼猪生长性能智能测定系统设计和试验[J].现代农业装备,2017(2):65-67.
[3]伍佰鑫,浣成,张翠永,等.规模猪场人工智能监控和预警的研究和应用概况[J].中国猪业,2017(12):60-61.
[4]雷彬,宋忠旭,孙华,等.不同饲养密度下猪的采食规律研究[J].养猪,2015(5):47-48.
[5]郁厚安.猪运动信息获取及分析[D].武汉:华中农业大学,2017.
[6]李颀,王丹聪.基于多传感器的猪只行为辨识[J].黑龙江畜牧兽医,2018(9):95-99.
[7]米国芹.猪群健康状态识别与掌握[J].中国畜禽种业,2018,14(9):137.
[8]张文文.基于模糊推理的多源信息生猪异常行为综合监测[D].太原:太原理工大学,2018.
[9]张振华,田建艳,王芳,等.猪咳嗽声特征参数提取与识别的研究[J].黑龙江畜牧兽医,2017(23):18-22.
[10]周丽萍,陈志,陈达,等.基于改进Otsu算法的生猪热红外图像耳根特征区域检测[J].农业机械学报,2016,47(4):228-232.
[11]谭辉磊,朱伟兴.基于轮廓的猪只饮水行为识别[J].江苏农业科学,2018,46(15):166-170
[12]赵伟,朱伟兴.基于Gabor方向直方图与猪体毛发模式特征的猪个体身份识别[J].江苏农业科学,2018,46(16):179-184.
[13]高云,郁厚安,雷明刚,等.基于头尾定位的群养猪运动轨迹追踪[J].农业工程学报,2017,33(2):220-226.
[14]王传哲,王东,张海辉,等.基于姿态角的生猪行为识别方法研究[J].扬州大学学报(农业与生命科学版),2016,37(4):43-48.
[15]肖德琴,冯爱晶,杨秋妹,等.基于视频追踪的猪只运动快速检测方法[J].农业机械学报,2016,47(10):351-357,331.