基于图像纹理特征的牛肉嫩度预测方法研究

Prediction of Beef Tenderness Based on Image Texture Features

在经过图像预处理,背最长肌与大理石花纹的分割,并实现大理石花纹特征值的提取后,利用灰度共生矩阵提取4个对嫩度剪切力贡献较大的纹理特征参数,并统计这些参数应用多元线性回归建立牛肉嫩度剪切力预测模型。结果表明:可见光下利用纹理特征预测牛肉嫩度的方法能够以96%的准确率实现嫩度剪切力等级的预测,具有较高的商用开发价值。

A mathematical modeling method for predicting beef tenderness utilizing image texture features under visible light was proposed.After image preprocessing,beef longissimus dorsi muscle and marbling were segmented,and then four marbling features that greatly influence beef shear force as a measure of meat tenderness were extracted by grey-level co-occurence matrix（GLCM） technique and statistically analyzed to establish a multiple linear regression model for predicting beef shear force.The proposed predictive method for beef shear force allowed 96% accurate prediction of beef tenderness,indicating its high value for commercial application.