基于图像特征的籽棉品级分级模型的研究

为了提高在籽棉收购环节中品级分级的客观性和准确性，籽棉品级分级模型以籽棉图像的特征参数为依据，以BP神经网络为分类算法。依据GB1103-2007中籽棉品级分级标准，提取籽棉图像的白度、黄度、杂质作为特征参数，使用部分样本对BP神经网络进行训练，训练后的BP神经网络对未参加训练的样本进行分级，精度达到81%。实验证明，该模型能够提高籽棉分级的客观性和准确性。

Grading model of seed cotton based on image features

In order to improve the accuracy and objectivity of classification of seed cotton, the paper presented a grading model. The model was based on image features and used BP neural networks as the classification algorithm. According to GB1103-2007 national standards of seed cotton, white degree, yellow degree and impurity degree were extracted as characteristic parameters. The BP neural networks were trained with part of the samples. After completion of the training, the left samples were substituted into the trained networks, and the result showed that classification precision of the model was 81%. The experimental results show that this model can promote grading accuracy and objectivity.