基于图像的羊体尺参数测点检测方法的研究  

Detection Method of Sheep Body Size Parameters Measurement Points Based on Image

作　　者：周艳青 薛河儒 白洁 乌丹牧其尔 马学磊 姜新华 ZHOU Yanqing;XUE Heru;BAi Jie;Wudanmuqier;MA Xuelei;JIANG Xinhua(College of Computer and Information Engineering,Inner Mongolia Agricultural University,Hohhot O10011,China)

机构地区：[1]内蒙农业大学计算机与信息工程学院,呼和浩特010011

出　　处：《内蒙古农业大学学报（自然科学版）》2024年第2期69-77,共9页Journal of Inner Mongolia Agricultural University(Natural Science Edition)

基　　金：国家自然科学基金项目(31960494);内蒙古自然科学基金项目(2020BS06003,2023QN06006);内蒙古农业大学基础学科科研启动项目(JC2019004)。

摘　　要：为了解决羊体尺参数测量中存在的测量困难、应激反应、人畜共患病等问题,寻找体尺参数测点显得尤为重要。试验基于图像对测点检测开展研究。首先,通过多尺度Retinex算法增强羊体图像;然后,利用图割算法识别羊体信息,并优化羊体轮廓;接着,提取羊体的包围盒,并划分为相等的4个区域,依据测点特征依次寻找7个测点;最后,基于测点计算羊体长、体高、臀高、胸深等4个参数。结果表明:体长、体高、臀高、胸深的相对误差分别为2.10%、2.05%、1.70%、0.81%,精度较高。这说明试验方法能够克服传统中存在的测量问题,实现零接触的羊体尺参数的获取,为羊生长发育及评估提供了可靠、快速和精准的信息。In order to solve the problems of measurement difficulties,stress response and zoonosis in the measurement of body size parameters of sheep,it is particularly important to find the measurement points of body size parameters.The test carried out a study on the detection of measuring points based on images.Firstly,the multi-scale Retinex algorithm was used to enhance the sheep im-age.Then,the graph cut algorithm was used to identify the information of sheep and optimize the contour of sheep body.Then,the surrounding box of sheep body was extracted and divided into four equal regions,and 7 measuring points were found according to the characteristics.Finally,the four parameters of sheep body length,body height,hip height and chest depth were calculated.The re-sults show that the relative errors of body length,body height,hip height and chest depth are 2.10%,2.05%,1.70%and 0.81%,respectively.The experimental method can overcome the traditional measurement problems and realize the acquisition of sheep body size parameters with non-contact,providing reliable,rapid and accurate information for sheep growth and development and evalua-tion.

关 键 词：羊体尺参数 多尺度RETINEX 图割算法 测点 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]