摘要
肌内脂肪(IMF)含量是衡量猪肉品质的重要指标,监测IMF的变化,有助于提高生猪养殖和育种质量。以42组Duroc生猪的B超图像为研究对象,根据屠宰测量得到的IMF含量反推图像分割阈值,然后计算8组与阈值相关的图像参数,利用多元线性回归分析进行参数筛选,得到预测阈值的最优拟合方程式,最后根据预测得到的分割阈值计算B超图像的IMF含量。试验结果表明,图像分割得到的IMF值与屠宰IMF值之间的相关系数为0.67,秩相关系数为0.69,经F测验差异显著。
The intramuscular fat (IMF) is an important indicator to measure the quality of pork. Monitoring the change of IMF contributes to swine farming and breeding. This research took 42 B-scan ultrasonic images of Duroc pigs as objects of study. First, a slice from the 10th to the 11 rib loin interface was used to determine carcass loin intramuscular fat percentage(CIMF) after harvest. Then CIMF were used to calculate the image segmentation threshold. Second, extracted 8 image parameters related to thresholds from B-scan images. And author built a model to predict new thresholds using multiple linear regression analysis with old thresholds as the dependent variable, 8 image parameters as independent variable. Finally, the new thresholds were used to predict loin intramuscular fat percentage (PIMF). The results showed that, the product moment coefficient and rank correlation coefficients between PIMF and CIMF were 0.67 and 0.69. After likelihood ratio test, the difference was significant.
出处
《广东农业科学》
CAS
CSCD
北大核心
2012年第21期128-131,F0002,共5页
Guangdong Agricultural Sciences
基金
国家科技支撑计划项目(2011BAD28B01)
广东省自然科学基金(10151064201000028)
关键词
肌内脂肪
B超图像
回归分析
intramuscular fat
ultrasound
regression analysis