基于生猪外形特征图像的瘦肉率估测方法
发布时间：2018-11-12 08:23
【摘要】：为实现生猪瘦肉率的快速无损检测,以机器视觉为主要技术,通过生猪的外形特征图像进行瘦肉率估测,为饲养者与收购者提供生猪品级的决策依据。采用MATLAB为开发工具,通过图形用户界面(graphical user interface,GUI)实现软件操作界面,以生猪的侧面及背面图像为研究对象,利用图像处理技术从目标中提取体长、体高、胸深、腹长、臀宽、腰宽等数据,以这些体尺的比例(胸深体高比、臀宽体长比、臀宽腰宽比、腹长体长比)为参数,通过径向基函数(radial basis function,RBF)神经网络进行瘦肉率估测。该文分别对7组生猪外形图像进行处理,4项比例指标的平均估测准确率分别为92.90%、92.44%、95.17%、96.51%,瘦肉率的平均估测准确率为94.35%。结果表明,该文所构造的基于生猪外形特征图像的瘦肉率估测方法工作效率高,成本低,可用于估测生猪瘦肉率。
[Abstract]:In order to realize the fast nondestructive testing of lean meat rate of live pigs, machine vision is used as the main technology to estimate the lean meat rate of live pigs by using the shape characteristic image of live pigs, which can provide the decision basis for pig quality grade for breeders and purchasers. Using MATLAB as the development tool, the software operation interface is realized through the graphical user interface (graphical user interface,GUI). The side and back images of the pig are taken as the research object, and the body length, body height, chest depth and abdomen length are extracted from the target by image processing technology. Based on the data of hip width, waist width and so on, the lean meat rate was estimated by radial basis function (radial basis function,RBF) neural network based on the ratio of breast depth to body height, hip width to body length, hip width to waist-width ratio and abdomen length to width ratio. In this paper, the contour images of 7 groups of live pigs were processed. The average accuracy of the four indexes was 92.90 and 95.179.179.51, respectively, and the average accuracy of the lean meat rate was 94.355.The average estimation accuracy of the four indexes was 92.90 and 95.179.51, respectively. The results show that the proposed method based on pig shape feature image has high efficiency and low cost and can be used to estimate lean meat rate of live pigs.
【作者单位】： 华南农业大学工程学院/教育部南方农业机械与装备关键技术重点实验室/广东省食品质量安全重点实验室;安徽省农业科学院农业经济与信息研究所;
【基金】：广东省科技计划项目(2012A020602039) 广州市产学研协同创新重大专项(201508010013)
【分类号】：TP391.41;TS251.51
本文编号：2326590
[Abstract]:In order to realize the fast nondestructive testing of lean meat rate of live pigs, machine vision is used as the main technology to estimate the lean meat rate of live pigs by using the shape characteristic image of live pigs, which can provide the decision basis for pig quality grade for breeders and purchasers. Using MATLAB as the development tool, the software operation interface is realized through the graphical user interface (graphical user interface,GUI). The side and back images of the pig are taken as the research object, and the body length, body height, chest depth and abdomen length are extracted from the target by image processing technology. Based on the data of hip width, waist width and so on, the lean meat rate was estimated by radial basis function (radial basis function,RBF) neural network based on the ratio of breast depth to body height, hip width to body length, hip width to waist-width ratio and abdomen length to width ratio. In this paper, the contour images of 7 groups of live pigs were processed. The average accuracy of the four indexes was 92.90 and 95.179.179.51, respectively, and the average accuracy of the lean meat rate was 94.355.The average estimation accuracy of the four indexes was 92.90 and 95.179.51, respectively. The results show that the proposed method based on pig shape feature image has high efficiency and low cost and can be used to estimate lean meat rate of live pigs.
【作者单位】： 华南农业大学工程学院/教育部南方农业机械与装备关键技术重点实验室/广东省食品质量安全重点实验室;安徽省农业科学院农业经济与信息研究所;
【基金】：广东省科技计划项目(2012A020602039) 广州市产学研协同创新重大专项(201508010013)
【分类号】：TP391.41;TS251.51
【相似文献】
相关期刊论文 前2条
1 王晓军;;中等径级杉树原木的制材——不同外形特征原木的下锯方法及其出材率[J];国外林业;1991年03期
2 ;[J];;年期
相关重要报纸文章 前2条
1 常丽君;基因位置对生物进化影响更大[N];科技日报;2010年
2 ;喂养动物和种植花草[N];中国教育资讯报;2002年
相关硕士学位论文 前1条
1 吴权石;基于知识的飞机外形特征参数化设计[D];南昌航空大学;2014年
，本文编号：2326590
本文链接：https://www.wllwen.com/kejilunwen/ruanjiangongchenglunwen/2326590.html
