基于声音的纸张柔软度检测方法研究

Study on the Method of Paper Softness Detection Based on Sound

针对当前纸张柔软度检测仪器,存在量程小、适用对象少的问题;提出一种根据纸张声音能量特征的差异,采用击点能量法和主成分分析法(PCA),实现对纸张柔软度进行分级的方法;击点能量法给出待测对象柔软度的参考值;对不同柔软度纸张声音信号分帧,求取每帧的能量,利用PCA获取特征互不相关的训练集,在此基础上建立特征矩阵,然后根据方差最小原则判断待测纸张样本的柔软度级别;实验表明,该方法能将同型号打印纸的柔软度分为5级,正确率达到100%。

Currently, the paper softness testing instruments have the problem that the range is small and the applicable object is little. A method for grading the softness of the paper was proposed, which is based on the differences of sound energy feature, hit--point energy and principal component analysis （PCA）. Firstly, Hit point energy law given the test object softness reference value, and framed audio signals for different softness of the paper and obtained the energy of each frame. Then the PCA was used to obtain characteristics unrelated training set and established the characteristic matrix. Finally, the softness levels of the paper was determined by the principle of minimum variance. Ex- periments showed that the paper softness could be divided into five clifferent levels, and the correct rated of 100%.