基于计算机视觉的冷却牛肉嫩度分析方法

分别运用大理石纹和灰度空间相关矩阵两种方法对含脂肪组织和不含脂肪组织冷却牛肉的图像特征进行了分析。在含脂肪组织牛肉图像中提取了能反映牛肉大理石纹结构分布和数量信息的特征量，在不含脂肪组织牛肉图像中提取出3个反映牛肉纹理粗细的特征量，并用线性回归方法分别建立了特征量与肉嫩度等级间的数量关系。研究表明，灰度空间相关矩阵法适于不含脂肪组织牛肉的嫩度评价，大理石纹方法适于含脂肪组织牛肉的嫩度评价。

Analysis Method of Tenderness of Chilled Beef Based on Computer Vision

In this paper, two methods of marbling score and grey space correlation matrix were used to analyze the image features of fat and no fat structures in chilled beef. From image of fat structure, the information on the distribution and quantity features of marbling score structure was gathered. For beef with no fat structure, three features representing thickness and thin of beef texture were inspected. The relationships between these features and tenderness grade were figured out by the multiple regression analysis. The results show that the marbling score method is suitable for the appraisement of tenderness of fat structure beef and the grey space correlation matrix is suitable for the appraisement of tenderness of no fat structure beef.