基于凹点分析法的粘连鸡体分割方法研究  

Research on the segmentation method of adhesive chicken body based on concave point analysis

作　　者：张宝全 陆辉山 王福杰 李明明 王馨宇 Zhang Baoquan;Lu Huishan;Wang Fujie;Li Mingming;Wang Xinyu(School of Mechanical Engineering,North University of China,Taiyuan,030051,China;School of Instrument and Electronics,North University of China,Taiyuan,030051,China)

机构地区：[1]中北大学机械工程学院,太原市030051 [2]中北大学仪器与电子学院,太原市030051

出　　处：《中国农机化学报》2021年第2期164-170,183,共8页Journal of Chinese Agricultural Mechanization

基　　金：国家重点研发计划项目(2016YFD0700202)。

摘　　要：针对鸡体之间存在相互粘连问题,以散养绿壳蛋鸡为研究对象,提出一种基于凹点分析法的粘连鸡体分割新方法,该方法在不同的颜色空间下使用最大类间方差法(Maximum Between-Class Variance,OTSU)结合形态学运算,对采集的视频图像进行图像预处理后,通过分析粘连区域的凸缺陷轮廓,使用正方形模板确定凹点位置,然后对凹点进行随机匹配确定正确的凹点匹配方式,最终实现粘连鸡体的分割。对不同数量的粘连鸡体分割试验结果表明,该方法实现了2~4只粘连鸡体的分割,平均分割准确率为92.8%,平均运行时间为2.817 s,并对比极限腐蚀结合凹点搜寻方法、分水岭分割方法,两种方法的平均分割准确率分别为63.4%、71.6%,该文方法提高粘连鸡体分割准确率;对真实养殖环境下粘连鸡群的分割试验结果表明,该方法对复杂粘连鸡群实现了较好的分割。该方法可以为后续监控鸡群个体健康状况提供技术支持。In order to solve the problem of adhesion between chicken bodies,a new method of adhesion chicken body segmentation based on concave point analysis was proposed in this paper.In this method,the image was preprocessed by using the maximum interclass variance method combined with morphological operation under the different color space.By analyzing the outline of the convex defect in the adhesion region,the square template was used to determine the position of the concave point,and then the concave point was randomly matched to determine the correct matching mode of the concave point,and finally the segmentation of the adherent chicken body was realized.The results of segmentation experiments on different numbers of adhesion chickens show that this method could achieve the segmentation of multiple adhesion chickens,with an average segmentation accuracy rate of 92.8%and an average running time of 2.817 s,and compared with the extreme erosion combined concave point seek method and watershed segmentation method,the correct segmentation rates of the two methods were 63.4%and 71.6%,respectively.This method was superior to the other two methods.The experiment on segmentation of adhesion chickens in a real breeding environment shows that the paper method could achieve better segmentation of low-density adhesion chickens.The method in this paper could provide technical support for the subsequent monitoring of the individual health status of the flock.

关 键 词：粘连分割 凹点分析 凸缺陷 OTSU 

分 类 号：S24:[农业科学—农业电气化与自动化]