基于三维视觉的荷斯坦奶牛中遗传育种个体的最优选择  

Individuals Optimal Selection in Holstein Cow Genetic and Breeding Based on 3D Vision

作　　者：朱坤华[1] 李骞[1] 武书彦[1] Zhu Kunhua;Li Qian;Wu ShuYan(College of Automation,Henan University of Animal Husbandry & Economy,Zhengzhou 450011,China)

机构地区：[1]河南牧业经济学院自动化学院,郑州450011

出　　处：《科技通报》2017年第2期52-55,共4页Bulletin of Science and Technology

基　　金：河南省高等学校重点科研项目(16A520014);河南省高等学校重点科研项目(15B520009);河南省科技厅重点科技攻关(152102210320);河南省教育技术装备与实践教育研究重点项目(GZS026);河南牧业经济学院科技创新团队(HUAHE2015006)

摘　　要：提出基于三维视觉对荷斯坦奶牛进行体型线性评定,对奶牛个体实现规范化的生产性能测定,择优选取优良奶牛育种。首先,利用摄像分别采集奶牛三个方向的整体图像,运用图像采集卡数字转化技术将图像输入计算机,利用改进的Canny算子图像边缘检测算法对获取的奶牛图像进行去噪处理,通过调整和选取合适的阈值达到抑制噪声的目的;其次,将处理后的图片运用Fisher线性鉴别分析方法,归类不同种类奶牛图片样本,对归类后的图片进行标准图像和特征点模板匹配;最后,在目标特征中识别出体型线状符合优秀标准的荷斯坦奶牛进行育种繁育后代。实验证明,利用图像识别技术对奶牛的体型线性鉴定有助于选育出奶效率高、身体健康及对于机械挤奶适应较好的优质牛群。The 3D vision based on linear evaluation of Holstein cows on dairy production performance, the individual to achieve standardized determination, select excellent dairy breeding. First of all, the overall image pickup were collected and cows in three directions, the use of image acquisition card digital conversion technology of images into a computer, the cow images are denoised by using Canny operator image edge detection algorithm, by adjusting and selecting appropriate threshold to achieve the purpose of noise suppression; secondly, after treatment the picture by Fisher linear discriminant analysis method,classified different pictures of cows samples, the classification of the picture after the standard image and feature template matching; finally, the target feature identifying body type linear meet the standards of excellence of the Holstein cow breeding breed. The experimental results show that the use of image recognition technology to identify the shape of the cow is conducive to the breeding of high efficiency,good health and good adaptability to mechanical milking cows.

关 键 词：三维视觉 荷斯坦奶牛 遗传育种 最优选择 

分 类 号：S823.91[农业科学—畜牧学] TP391[农业科学—畜牧兽医]