基于最小外接矩形的稻米粒型检测方法

A Study on Inspection of Rice Kernel Ratio with Minimum Enclosing Rectangle Method

粒型是稻米分级的重要特征之一。对利用计算机图像处理技术检测稻米粒型的方法进行了探索,设计了一种利用稻米轮廓的最小外接矩形(MER,MinimumEnclosingRectangle)计算粒型的方法,并对算法进行了验证。实际测量值与计算值的相关性为0.959,最大误差在0.21以内,最大相对误差为5.69%,平均相对误差为3.21%。最小外接矩形计算方法不仅为进一步研究开发基于图像处理技术的稻米外观品质检测系统打下了基础,而且可用于对其他谷物进行粒型检测。

The ratio of rice kernel is one of the most important features in classification of rice kernel. A method to inspect the ratio of rice kernel was proposed, in which a calculation method by using the minimum enclosing rectangle （MER） around the rice kernel was designed and verified. The experimental result showed that the correlation coefficient of real ration versus detected one reached 0. 959, the maximum error was below 0.21, the maximum relative error was 5.69 % , and the average relative error was 3.21%. This method could not only lay a foundation for further developing rice quality detection system by using image processing technology, but also be used to detect the ratio of other grains.