机器视觉的淡水鱼重量预测
1为便于淡水鱼后续加工,需对其进行重量预测,在我国它的加工处理还主要依靠人力。人工称量鱼体重量不仅作业环境差,且对鱼体有损伤,针对这些问题,本文以常见淡水鱼鲫鱼为研究对象,采用机器视觉和数字图像处理技术实现鲫鱼重量的自动预测。首先提取了鲫鱼的纹理特征和形状特征,并对研究对象进行了基于纹理和形状的两次分割,精确分割鱼头、身体、尾巴三个部分。然后获取各个部分的投影面积,运用各部分占总质量的比例对投影面积进行校正,再利用回归分析建立重量预测模型。本文用20条鲫鱼作为建模样本,7条作为测试样本,用预测重量与实际重量进行比较,得出平均预测误差仅为4.63%,并对误差原因进行了分析。
目 录
Abstract1
Key words1
引言1
1 机器视觉技术2
1.1 研究目的及意义2
1.2 机器视觉技术简介及应用现状2
1.2.1 机器视觉技术简介2
1.2.2 机器视觉技术应用现状2
1.3 研究内容与技术路线3
1.3.1 主要研究内容3
1.3.2 研究技术路线3
2 图像信息采集及图像处理4
2.1 图像信息采集4
2.2 图像背景去除4
2.2.1 图像裁剪5
2.2.2 图像阈值分割与边缘检测5
2.2.3 图像数学形态操作6
2.2.4 背景去除8
3 纹理、形状特征提取与分析 8
3.1 纹理特征提取8
3.1.1 图像灰度化8
3.1.2 灰度共生矩阵9
3.1.3 灰度共生矩阵的特征参数9
3.2 纹理特征参数分析10
3.2.1 贝叶斯分类器10
3.2.2 最优特征提取12
3.3 基于纹理分割12
3.4 形状特征提取^15
3.4.1 长短轴15
3.4.2 定位鱼头、鱼尾位置15
3.5基于形状特征的分割20
4 重量预测 21
4.1 图像特征提 *好棒文|www.hbsrm.com +Q: ^351916072^
取21
4.2 重量预测模型建立23
4.3 模型检验23
4.4模型分析、讨论与设想24
4.4.1 模型分析、讨论24
4.4.2 设想24
5 总结25
致谢27
参考文献27
基于机器视觉的淡水鱼重量预测
计算机科学与技术专业学生 侯思宇
指导教师 谢忠红
Research on weight prediction based on machine vision
Student majoring in computer science and technology Hou siyu
Tutor xie zhonghong
Abstract: In order to facilitate the subsequent processing, it is necessary to develop a grading system for weight prediction of freshwater fish automatically. At present, it does still mainly rely on the processing of human in China. However, rely on manual weigh, not only poor operating environment, but there is damage to the fish. In view of this problem, the usual freshwater fish carp were selected as the objects for study, researches the weight prediction by using machine vision technology and digital image. For one thing, extracted the carp’s texture features and shape features and as well conducted two division bases on texture and shape on object. Precise division head, body, tail three parts, obtain accurate data about projected area of those three parts. For another, calculated the percentage about head、body、tail occupy global weight, which were used to correct the projected area. Finally the weight prediction model was extracted by the regression analysis. In this paper, 20 samples of crap as a model, seven as a test sample, and the forecast model was verified. The mean relative error was 4.63%, and then the paper analyzes the causes of errors.
Key words: machine vision;freshwater fish;regression analysis ;Weight prediction
引言
我国是世界淡水鱼养殖、生产大国,淡水鱼作为主要动物蛋白质来源之一,在我国人民的食物结构中占有重要的位置[1]。但目前我国淡水鱼的加工处理工业机械化程度低、技术含量不高,在很大程度上阻碍了我国淡水渔业的发展[2]。加工处理中前处理是保证淡水鱼加工质量及其商品价值的重要环节,且鱼体重量称量是前处理的重要工序之一。部分产业采用的机械化称重,对鱼体损伤比较严重。因此为了提高淡水鱼的附加价值、扩大淡水鱼市场,基于机器视觉的淡水鱼重量预测势必要成为淡水鱼前处理环节的必要要求。目前基于机器视觉对于农产品的品种识别和重量预测是一个研究热点。国内外也做出了一些相关研究内容。
目 录
Abstract1
Key words1
引言1
1 机器视觉技术2
1.1 研究目的及意义2
1.2 机器视觉技术简介及应用现状2
1.2.1 机器视觉技术简介2
1.2.2 机器视觉技术应用现状2
1.3 研究内容与技术路线3
1.3.1 主要研究内容3
1.3.2 研究技术路线3
2 图像信息采集及图像处理4
2.1 图像信息采集4
2.2 图像背景去除4
2.2.1 图像裁剪5
2.2.2 图像阈值分割与边缘检测5
2.2.3 图像数学形态操作6
2.2.4 背景去除8
3 纹理、形状特征提取与分析 8
3.1 纹理特征提取8
3.1.1 图像灰度化8
3.1.2 灰度共生矩阵9
3.1.3 灰度共生矩阵的特征参数9
3.2 纹理特征参数分析10
3.2.1 贝叶斯分类器10
3.2.2 最优特征提取12
3.3 基于纹理分割12
3.4 形状特征提取^15
3.4.1 长短轴15
3.4.2 定位鱼头、鱼尾位置15
3.5基于形状特征的分割20
4 重量预测 21
4.1 图像特征提 *好棒文|www.hbsrm.com +Q: ^351916072^
取21
4.2 重量预测模型建立23
4.3 模型检验23
4.4模型分析、讨论与设想24
4.4.1 模型分析、讨论24
4.4.2 设想24
5 总结25
致谢27
参考文献27
基于机器视觉的淡水鱼重量预测
计算机科学与技术专业学生 侯思宇
指导教师 谢忠红
Research on weight prediction based on machine vision
Student majoring in computer science and technology Hou siyu
Tutor xie zhonghong
Abstract: In order to facilitate the subsequent processing, it is necessary to develop a grading system for weight prediction of freshwater fish automatically. At present, it does still mainly rely on the processing of human in China. However, rely on manual weigh, not only poor operating environment, but there is damage to the fish. In view of this problem, the usual freshwater fish carp were selected as the objects for study, researches the weight prediction by using machine vision technology and digital image. For one thing, extracted the carp’s texture features and shape features and as well conducted two division bases on texture and shape on object. Precise division head, body, tail three parts, obtain accurate data about projected area of those three parts. For another, calculated the percentage about head、body、tail occupy global weight, which were used to correct the projected area. Finally the weight prediction model was extracted by the regression analysis. In this paper, 20 samples of crap as a model, seven as a test sample, and the forecast model was verified. The mean relative error was 4.63%, and then the paper analyzes the causes of errors.
Key words: machine vision;freshwater fish;regression analysis ;Weight prediction
引言
我国是世界淡水鱼养殖、生产大国,淡水鱼作为主要动物蛋白质来源之一,在我国人民的食物结构中占有重要的位置[1]。但目前我国淡水鱼的加工处理工业机械化程度低、技术含量不高,在很大程度上阻碍了我国淡水渔业的发展[2]。加工处理中前处理是保证淡水鱼加工质量及其商品价值的重要环节,且鱼体重量称量是前处理的重要工序之一。部分产业采用的机械化称重,对鱼体损伤比较严重。因此为了提高淡水鱼的附加价值、扩大淡水鱼市场,基于机器视觉的淡水鱼重量预测势必要成为淡水鱼前处理环节的必要要求。目前基于机器视觉对于农产品的品种识别和重量预测是一个研究热点。国内外也做出了一些相关研究内容。
版权保护: 本文由 hbsrm.com编辑,转载请保留链接: www.hbsrm.com/jsj/jsjkxyjs/1867.html