数字图像和逐步回归客观评定冷却猪肉肉色

Objective evaluation of chilled pork color by digital image and stepwise regression

为开发准确、快速的猪肉肉色质量客观评定方法,研究了数字图像处理和和逐步回归模型对冷却猪肉肉色客观评定分级的效果。对宰后冷却24h的猪胴体,切开第3～4肋骨间背最长肌,发色60min,数码相机获取数字图像处理后提取断面肉色参数(L*、a*、b*、Chroma、Hueangle)。提取的80头猪胴体背最长肌肉色参数经逐步回归建立了肉色评定模型。结果表明,数字图像处理后提取肉色参数建立的逐步回归模型评定冷却猪肉肉色分值的效果优于BP人工神经网络模型;若以∣评定肉色分值-感官肉色分值∣≤0.3为评定正确判断标准,前者评定正确率为78.8%,后者为60.4%;前者与本试验评定正确率最高的单个感官评定人员相比(78.2%),差异不显著(P>0.05)。因此,数字图像处理可有效地对冷却猪肉肉色进行客观评定。

To develop accurate and rapid method for objective evaluation, the effect of digital image processing and stepwise regression on chilled pork color evaluation were studied. Pork carcass longissimus dorsi muscles were cut at 3～4th rib and bloomed for 60min. Digital images of the muscle surface were captured by digital camera and processed to extract image color features（L^＊, a^＊, b^＊, Chroma, Hue angle）. Extracted color features from 80 carcass longissimus dorsi muscles were used to establish color score evaluation model by stepwise regression . The results showed the color score evaluation model by stepwise regression based on color features from digital image processing was better than that by BP artificial neural network and the evaluation accuracy was 78.8% and 60.4%, respectively if judged by the formula∣evaluation color score-sensory color score∣≤0.3. The former had no significant variation with evaluation accuracy（78.2%） compared with the best single panelist （P〉0.05）. Therefore, digital image processing is an effective tool for objective evaluation of chilled pork color.