为实现合格和缺陷板栗的分级， 研究了 1 种基于 BP 神经网络与板栗图像特征的板栗分级方法。 试验以罗田板

栗为研究对象， 提取的颜色及纹理等 8 个特征值， 通过主成分分析提取相应的主成分得分向量构成模式识别的输入。 利

用 BP 神经网络方法建立了板栗分级模型。 试验结果表明， 在图像信息主成分因子数为 3， 中间层节点数为 12 时， 建立

的模型最佳， 模型训练时的回判率为 100 ， 预测时识别率达到了 91 .67 。 研究结果表明基于机器视觉技术的针对缺陷

板栗分级检测方法是可行的。- In order to realize grading of eligible and defected chestnut by using machine vision, a classification method

of chestnut was developed based on BP-ANN and image feature of chestnut. In this experiment, Luotian chestnuts were

used as experimental targets. Principal component analysis (PCA) was implemented on these feature variables from

eight eigen values including color parameters and veins characteristics parameters etc., and principal components (PCs)

vectors were extracted as the inputs of pattern recognition. Grading models were built by BP neural network. The test

result showed that when the number of principal component factor was three and the number of nodes of hidden layer

was twelve, the discriminating rate was as high as 100　in training set, and 91.67　in prediction set. The overall results

shows that it is feasible to discriminate chestnut quality with machine vision.