Agriculture and Agri-Food Canada, Lacombe Research Centre, Lacombe, Alberta T4L 1W1, Canada.
Agriculture and Agri-Food Canada, Lacombe Research Centre, Lacombe, Alberta T4L 1W1, Canada.
Meat Sci. 2021 Sep;179:108549. doi: 10.1016/j.meatsci.2021.108549. Epub 2021 May 7.
An image analysis method for identifying marbling on exposed boneless pork loin surfaces was developed, solving the problem of influences from uneven lean colour. Comparing results from RAW and JPEG images of the mid-loin pork chop showed that, except at very high marbling levels, estimation of %IMF (intramuscular fat) was similar with both image types (R = 0.72 and 0.74 respectively). Marbling calculated from JPEGs was positively related to %IMF, but also diverged as %IMF increased. Estimations of %IMF at the mid-loin chop from JPEGs of external loin sites showed marbling on the posterior face to be the most strongly related (r = 0.72, P < 0.0001), followed by the ventral surface (r = 0.62-0.64, P < 0.0001), and the anterior face (r = 0.56, P < 0.0001). For on-line application, computing speed is important. This study suggested that an image capture system could be reduced to the capture of a cyan-filtered monochrome JPEG image for analysis.
一种用于识别暴露无骨猪里脊表面大理石花纹的图像分析方法已经开发出来，解决了由于瘦肉颜色不均匀而产生的影响。比较中腰猪排的 RAW 和 JPEG 图像的结果表明，除了非常高的大理石花纹水平外，两种图像类型的肌内脂肪（IMF）估计值相似（分别为 R = 0.72 和 0.74）。从 JPEG 计算得出的大理石花纹与 IMF%呈正相关，但随着 IMF%的增加也会出现差异。从外部腰肉部位的 JPEG 对中腰猪排的 IMF%进行估计表明，后腹面的大理石花纹与 IMF%的相关性最强（r = 0.72，P < 0.0001），其次是腹侧表面（r = 0.62-0.64，P < 0.0001），以及前腹侧（r = 0.56，P < 0.0001）。对于在线应用，计算速度很重要。本研究表明，图像采集系统可以简化为采集一个经过青色滤光片的单色 JPEG 图像进行分析。
