摘要
为了提高苹果分级的准确率和稳定性,在图像处理的基础上,基于Fourier描述子和HIS颜色模型分别提取了苹果的形状和颜色两类主要外观特征,并分别用神经网络进行单特征初步分级,将其结果作为证据,通过D-S证据理论进行决策级融合,根据分类阈值得到最终分级结果。实验结果表明,该方法分级正确率达93.75%,与单指标特征分级相比,识别率高,稳定性好。
In order to increase the accuracy and stability of apple gradings,hape and color features which can show the ap-ples’ appearance quality are separately extracted by Fourier descriptor and HIS color model.Firstlyt,he apples are graded re-spectively by neural network.Thent,he former grading results are used as evidences to achieve the decision fusion.Finally,us-ing identification threshold to get the grades.The experimental results show that the grading accuracy reaches 93.75%t,he pro-posed method has good performance on accuracy and stability compared to the grading method based on single feature.
出处
《计算机工程与应用》
CSCD
北大核心
2010年第35期202-204,234,共4页
Computer Engineering and Applications
基金
盐城工学院重点建设学科开放基金(No.XKY2010021)