基于机器视觉的奶牛体尺参数测量研究

发布时间：2019-06-22 17:34

【摘要】：奶牛的体尺参数是评价奶牛成长状况中一项非常重要的参数,包含奶牛体高、体斜长、尻角度、乳头长度等参数,对奶牛养殖具有重要意义。传统的奶牛体尺数据是由人工直接进行接触式测量获得的,这种测量方式工作量大、测量条件相对艰苦、影响奶牛正常的生活习性。机器视觉中的视觉测量技术具有与被观测的对象不进行接触、不会出现视觉疲劳现象、测量精度高、稳定性强等优点,将视觉测量技术应用于奶牛体尺测量,把机器视觉技术应用于奶牛畜牧养殖领域,加快实现奶牛的精细化饲喂养殖。本课题结合摄像机标定、轮廓提取、SIFT特征点匹配等图像处理技术,将机器视觉应用于奶牛体尺测量,其目的在于获取奶牛的体尺数据,实现奶牛体尺的视觉测量。主要研究内容如下:(1)对传统摄像机标定进行详细的叙述,并建立摄像机成像模型;介绍了双目视觉测量的原理,并对视觉测量建立了数学模型;介绍了采用平面棋盘格为标定物的张正友标定法,针对选取的软件设计制作出77?的圆形标记点的标定板,通过单目和双目的标定试验,得到摄像机的内外相机参数。(2)对视觉测量系统进行介绍,针对测量系统的需要进行摄像机类型、镜头以及图像采集卡等硬件的选型,课题选择具有强大图像处理功能的HALCON软件,针对课题研究进行奶牛轮廓特征的提取。(3)介绍了SIFT特征点匹配算法,首先通过对图像进行多尺度空间极值点的检测,其次获取特征关键点并对关键点进行精确定位分析,然后计算出关键点的主方向,最终实现对描述子的构造,得到图像特征的特征向量。其中对图像进行多尺度空间极值点的检测和关键点的精确定位的目的在于实现在DOG尺度空间中提取极值点并进行优化从而获取特征点。通过初步试验验证,本测量系统可以实现奶牛体尺的测量,并且具有工作稳定、测量精度高等特点,为进一步深入研究奶牛的体型与体况,实现数字化养殖打下了基础,具有很好的市场应用前景。
[Abstract]:The body size parameters of dairy cows are a very important parameter in evaluating the growth of dairy cows, including cow body height, oblique length, rump angle, nipple length and so on, which is of great significance to cow breeding. The traditional body size data of dairy cows are obtained by manual direct contact measurement. This kind of measurement method has a large workload and relatively difficult measuring conditions, which affects the normal living habits of dairy cows. The visual measurement technology in machine vision has the advantages of not contact with the observed object, no visual fatigue phenomenon, high measurement accuracy and strong stability. The visual measurement technology is applied to the measurement of dairy cow body size, and the machine vision technology is applied to the field of dairy cattle breeding, so as to speed up the realization of fine feeding and breeding of dairy cows. Combined with camera calibration, contour extraction, SIFT feature point matching and other image processing techniques, machine vision is applied to cow body size measurement, the purpose of which is to obtain the body size data of dairy cows and realize the visual measurement of cow body size. The main research contents are as follows: (1) the traditional camera calibration is described in detail, and the camera imaging model is established; the principle of binocular vision measurement is introduced, and the mathematical model of visual measurement is established. Zhang Zhengyou calibration method, which uses plane chessboard as calibration object, is designed and made for the selected software. The calibration board of circular marking point is used to obtain the internal and external camera parameters of the camera through monocular and binocular calibration tests. (2) the visual measurement system is introduced. According to the needs of the measurement system, the camera type, lens, image acquisition card and other hardware are selected, and the HALCON software with powerful image processing function is selected. Aiming at the research of dairy cow outline feature extraction. (3) the SIFT feature point matching algorithm is introduced. Firstly, the multi-scale spatial extreme point detection of the image is carried out, then the feature key points are obtained and the key points are accurately located and analyzed, then the main direction of the key points is calculated, and finally the construction of the descriptive points is realized, and the feature vector of the image features is obtained. The purpose of detecting the extreme points and accurately locating the key points in the multi-scale space is to extract the extreme points in the DOG scale space and optimize them to obtain the feature points. The preliminary test shows that the measurement system can realize the measurement of dairy cow body size, and has the characteristics of stable work and high measurement accuracy, which lays a foundation for further study of cow body size and body condition and the realization of digital breeding, and has a good market application prospect.
【学位授予单位】：山东农业大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：S823;S818.9

【相似文献】