基于计算机视觉的午餐肉颜色测定系统研究

目的建立一种对午餐肉样品物理特性要求较少,能对物料表面整体颜色进行准确测量的无损检测方法。方法采用计算机视觉系统对24色色彩测试板测定得L,a,b值,使用色彩色差计对24色色彩测试板测得L~*,a~*,b~*值,对两组数据进行线性回归;计算机视觉系统测定午餐肉的L,a,b值,带入回归方程得到标准的L,a,b值,色彩色差计对午餐肉测定得L~*,a~*,b~*值,用SPSS软件对得到的标准L,a,b值和L~*,a~*,b~*值进行成对样本检验。结果 L,a,b值回归方程的相关系数r~2分别为0.9900、0.9707和0.9801,有高度相关性;午餐肉标准L,a,b值和L~*,a~*,b~*值成对样本检验得到的P值分别为0.146、0.087、0.109,大于显著性水平0.05,回归方程转换值与色差计测定结果无显著差异。结论本文建立的基于计算机视觉的午餐肉颜色测定方法可以准确测定午餐肉颜色,其效果可以代替色差计。

Study on color detecting system of luncheon meat based on computer vision

Objective To provide a nondestructive testing method with less requirement in physical characteristics of tested sample, which could measure the whole surface color of food accurately. Methods The data of L, a and b of 24 color test boards were measured by computer vision and color difference meter (CDM). Linear regression was done by the two groups of data. The two groups of data went through a paired-sample-test implemented by SPSS software. Results The correlation coefficient of L, a and b measured by computer vision system (CVS) and CDM were 0.9900, 0.9707 and 0.9801, respectively, which were highly correlated. P values of L, a and b through a paired-sample-test implemented by SPSS were 0.146, 0.087, and 0.109, respectively, which were bigger than the significant level 0.05. Thus these two methods had no obvious difference. Conclusion The CVS designed in this research can test the L, a and b values of luncheon meat accurately. CVS can replace the CDM for color detecting.