摘要
为了提高猪胴体分级的准确性,利用计算机视觉技术、图像处理技术及统计分析方法,对已建立的猪胴体分级标准及预测方程进行修订。结果表明:以左半胴体质量、臀中肌横长和臀中肌膘厚预测瘦肉率绝对误差小于4%;同时以瘦肉率、臀中肌膘厚、1/2横长处膘厚及6～7肋处膘厚等特征作为分级主要参数,使分级准确率达90%。将各处膘厚与瘦肉率相结合,并对猪胴体级别根据实际需求进行调整,可使分级工作更加合理,准确性也有提高。
To increase the accuracy of pig carcass grading, computer vision technology, image processing technology and statistical methods were used to modify the established pig carcass grading standard and prediction equations. An absolute error smaller than 4% was obtained from lean percentage predications based on half carcass weight, gluteus medium length and gluteus medium fat thickness. The accuracy of carcass grading obtained using lean meat percentage, gluteus medium fat thickness, mid-body fat thickness and rib 6--7 fat thickness as evaluation parameters was 90%. In conclusion, more reasonable and more accurate carcass grading can be achieved when using fat thickness in different carcass parts and lean percentage as evaluation parameters and making practical modifications to the carcass grades.
出处
《肉类研究》
2013年第2期1-4,共4页
Meat Research
基金
"十二五"国家科技支撑计划项目(2012BAK17B09
2012BAD28B02)
国家生猪产业体系北京市创新团队项目
关键词
猪胴体
计算机视觉
分级标准
瘦肉率
膘厚
pig carcass
computer vision
grading standard
lean meat percentage
fat thickness
