目的 解决包装材料复杂颜色区域色差评价，目测检验方法存在判定标准复杂、结果一致性差等问题。方法 提出一种基于包装材料高清数字图像的RGB颜色空间色差评价方法。通过搭建高清数字图像采集装置，获取包装材料标准样(上限、中限、下限)的全幅数字图像。通过采用RGB颜色空间降维的方法，对标准样测试区域的颜色空间进行离散边界拟合，构建标准样品测试区域的三限样RGB色差模型，与测试样对应测试区域的颜色空间进行对比;根据颜色空间模型的一致性判定测试样色差是否合格。结果 对比多种离散边界拟合模型，采用滚球法边缘拟合模型的包含率和多出率分别达到100%、78.9%，模型拟合效果最佳。通过对包装材料色差缺陷样验证测试，采用上述色差评价方法准确率达到100%。结论 本文提出的滚球法颜色空间边缘拟合模型和色差评价方法，可实现对包装材料复杂颜色区域的色差合格性判定，有利于提高企业生产的智能化检测水平。

The work aims to solve the problems of color difference evaluation of complex color areas of packaging materials, complex criteria and poor consistency of results in visual inspection methods. In this paper, an RGB color space color difference evaluation method based on high-precision digital images of packaging materials was proposed. By setting up a high-precision digital image acquisition device, the full digital image of the standard sample of packaging materials (upper limit, middle limit and lower limit) was obtained. By using the method of RGB color space dimensionality reduction, the color space of the test area of the standard sample was fitted with discrete boundary, and the three-limit sample RGB color difference model of the test area of the standard sample was constructed, and the color space of the corresponding test area of the test sample was compared. The color difference qualification of the test sample was judged according to the consistency of the color space model. Compared with various discrete boundary fitting models, the inclusion rate and excess rate of the rolling ball method edge fitting model reached 100% and 78.9%, respectively, and the model fitting effect was the best. Through the verification test of color difference defect samples of packaging materials, the accuracy of the above color difference evaluation method reached 100%. The rolling sphere color space edge fitting model and color difference evaluation method proposed in this paper can realize the color difference qualification judgment of complex color areas of packaging materials, which is conducive to improving the intelligent detection level of enterprise production.