摘要
借助于图像处理技术,提取缝纫平整度照片的图像灰度标准差、图像熵、小波变换系数标准差、小波信息熵等特征参数,建立了缝纫平整度的客观评判的概率神经网络模型。经过训练和检验,得出该模型的预测值与期望值之间的相关系数在0.99以上,说明网络模型有效,且精度高,可以用于预测未知缝纫样本的缝纫平整度等级。
In this paper,the standard deviation of image grey,Image entropy,standard deviation of wavelet coefficient and wavelet entropy about the photograph of seam pucker are extracted based on the image process technology.Then probabilistic neural network（PNN） model of evaluation of seam pucker is established.After training and testing,the correlation coefficient between prediction value and the expected value is above 0.99,which indicates that the network model has high precision and can be used to predict seam pucker of unknown samples.
出处
《丝绸》
CAS
北大核心
2011年第4期28-31,共4页
Journal of Silk
基金
上海市教委科技创新基金项目(10YZ170)
关键词
缝纫平整度
客观评判
图像处理
小波分析
概率神经网络
Seam pucker
Objective evaluation
Image process
Wavelet analysis
Probabilistic neural network
