基于等效椭圆和BP神经网络的马铃薯形状分类研究

Potato shape classification based on equivalent ellipse and BP neural network

根据国家马铃薯分级标准的要求,提出了一种基于区域的等效椭圆和BP神经网络相结合的马铃薯形状分类方法.首先运用等效椭圆来提取一组特征参数R和C,然后将这些特征参数输入到已训练好的BP神经网络完成对马铃薯的形状分类.结果表明:该方法选用的特征参数少,能较为有效的描述马铃薯的形状,分级结果准确率达94.7%,与人工分级的一致性高,能满足实际检测的要求.

The shape is one of the important features for potato integrated classification.According to the requirements of classification,a potato shape classification method combining region-based equivalent ellipse with BP neural network was presented.First,shape features were extracted from region-based equivalent ellipse,and then these features were input to the trained BP neural network to be completed the potato shape classification.The experiments showed that this method could describe the shape feature of the potato effectively by using the less feature parameters.The grading results were consistent with those of artificial classification highly,its precision was up to 94.7% and could meet the requirements of practical application.