Research automatic apples grading. Single information source apple classification can not fully reflect the quality of apple grading accuracy rate, and the accuracy is low and unstable. In order to improve apple grading accuracy rate, the paper proposed an apple grading method based on the evidence theory fusion. Firstly, image processing technology was used to extract the size, shape and color features to describe the quality of apple, then RBF neural network was used for single feature for preliminary classification. The preliminary classification results were taken as evidences, and the evidence theory was used for the fusion of preliminary classification results to achieve apple apple automatic grading grading. Simulation results show that the feature fusion classification method can improve the correct rate of apple automatic grading. Compared with a single feature classification, the grading accuracy rate is higher, ansthe stability is better.%研究苹果优化自动分级控制问题,苹果等级由多种特征共同决定,单一苹果特征分类算法不能全面反映苹果品质,导致分级正确率低.为提高提高苹果分级正确率,提出一种基于证据理论的特征融合苹果分级方法.首先提取大小、形状、颜色特征,然后采用RBF神经网络对每一个特征进行初步分级,将单一特征初步分级结果作为证据,最后采用证据理论对初步分级结果进行决策级融合,获得苹果多特征融合分级结果.仿真表明,相对于单一特征分级方法,特征融合分级方法提高苹果自动分级正确率,稳定性更好,是一种有效苹果自动分级方法.
展开▼
