基于图像传感技术的娃娃菜外观品质检测

Detection of external quality for baby cabbage by image sensing technology

中文关键词:  娃娃菜  机器视觉  分级  瑕疵识别  图像处理

英文关键词:baby cabbage  machine vision  grade  defect identification  image processing

中文摘要:

      目的 设计一套基于图像传感技术分析娃娃菜外观品质如尺寸、重量、瑕疵点等的检测方法。方法 搭建一套图像采集平台拍摄娃娃菜不同侧面, 应用图像处理技术, 分割出娃娃菜图像区域, 并数字化其区域特征信息(包括: 投影面积、尺寸、瑕疵点面积等)。结果 建立合格娃娃菜的侧面投影面积与重量真实值间的线性关系, 其相关系数为0.938, 均方根误差为36.52 g; 对比人工检测, 图像法可以识别出娃娃菜外表面95%的瑕疵点(腐黑点、裂纹裂缝等); 参照娃娃菜的分级标准, 以图像法获取各特征指标, 结合聚类算法分级娃娃菜, 其中K-medoid法准确率为100%, Gath-Geva法准确率为96.67%。结论 机器视觉技术可应用于娃娃菜的自动检测和分级, 为在线无损检测提供参考。

英文摘要:

      Objective To design a set of detection methods to analyze the appearance quality of baby cabbage such as size, weight, defects and so on based on image sensing technology. Methods A set of image acquisition platform was set up to capture the different sides of the baby cabbage. The image processing technology was applied to segment the image area of the baby cabbage and the regional feature information was digitized (including: projection area, size, defect area, etc.). Results The linear relationship between the side projection area of the qualified baby cabbage and the actual weight value was established, and the correlation coefficient was 0.938, and the root mean square error was 36.52 g. Compared with the manual detection, 95% of the defective points (rotten black spots, cracks and cracks) on the outer surface of the baby cabbage could be identified by the image method; with reference to the classification standard of baby cabbage, each characteristic index was obtained by image method, and baby cabbage was classified by combining clustering algorithm, among them, the accuracy rate of K-medoid method was 100%, and the accuracy rate of Gath-Geva method was 96.67%. Conclusion The machine vision technology can be applied to the automatic detection and grading of the baby cabbage, and provides a reference for online nondestructive detection.