引用本文
  •    [点击复制]
  •    [点击复制]
【打印本页】 【下载PDF全文】 查看/发表评论下载PDF阅读器关闭

←前一篇|后一篇→

过刊浏览    高级检索

本文已被:浏览 596次   下载 1259 本文二维码信息
码上扫一扫!
基于多尺度Retinex图像增强的羊体尺参数无接触测量
周艳青1, 薛河儒1, 姜新华1, 郜晓晶1, 杜雅娟1,2, 白明月1
0
(1.内蒙农业大学 计算机与信息工程学院, 呼和浩特 010018;2.内蒙古工业大学 数据科学与应用学院, 呼和浩特 010051)
摘要:
针对传统羊体尺参数测量中存在工作量大、精度低、应激反应强等问题,采用多尺度Retinex算法、Graph Cut算法和羊体尺测点识别相结合的方法,基于双目视觉检测原理对羊体尺参数的无接触测量进行研究。结果表明:1)带色彩恢复的多尺度Retinex算法能增强光照不均匀的羊图像,对羊图像的细节和颜色恢复表现出较强的处理能力;2)基于多尺度分水岭的Graph Cut算法准确地分割出羊体区域,获得光滑的羊体轮廓线;3)划分羊体轮廓线区域,采用包络线分析方法识别体尺测点,计算羊体尺参数,并与真实值比较,11只羊的平均相对误差为2.32%,除去绒山羊剩下9只羊的平均相对误差为1.95%,测量精度较高。试验证明本研究带色彩恢复的多尺度Retinex算法能增强羊图像亮度和色度,包络线分析方法能准确地识别体尺测点,算法稳定,能够实现饲养过程中羊体尺参数的无接触测量。
关键词:    无接触测量  体尺参数  多尺度Retinex  双目立体视觉
DOI:10.11841/j.issn.1007-4333.2018.09.19
投稿时间:2017-12-05
基金项目:国家自然科学基金项目(61461041);内蒙古自治区博士研究生科研创新项目(B20161012911)
Non-contact measurement of sheep body size based on multi-scale Retinex image enhancement
ZHOU Yanqing1, XUE Heru1, JIANG Xinhua1, GAO Xiaojing1, DU Yajuan1,2, BAI Mingyue1
(1.College of Computer and Information Engineering, Inner Mongolia Agricultural University, Hohhot 010018, China;2.School of Data Science and Application, Inner Mongolia University of Technology, Hohhot 010051, China)
Abstract:
Due to the problems of large workload, low accuracy and strong stress reaction existed in traditional sheep body size parameters measurements, on the basis of the principle of binocular vision detection, a non-contact measurement study for sheep body size parameters was implemented by the multi-scale Retinex algorithm, Graph cut algorithm and sheep body size measurement points extraction. The results showed that:1) The multi-scale Retinex algorithm with color restoration enhanced the image of sheep with uneven illumination, which displayed strong processing ability for the details and color restoration of sheep images. 2) The Graph Cut algorithm based on multi-scale watershed accurately segmented the sheep body region and obtained smooth sheep body contour line. 3) The sheep contour line was divided and envelopment analysis method was applied to identify the body size measurement points. The body size parameters were detected. Compared with the real values, the average relative errors for 11 sheep and 9 sheep without cashmere goats were 2.32% and 1.95%, respectively. The measurement accuracy proved that the multi-scale Retinex algorithm with color restoration could enhance the luminance and chrominance of sheep image and the envelopment analysis algorithm could accurately identify the measurement points of body size. The algorithms were stable, which could realize sheep body size detection contactless in sheep farm.
Key words:  sheep  non-contact measurement  body size parameter  multi-scale Retinex  binocular stereo vision