应用自适应正交小波对纺织品纹理信息表征的研究

项目名称： 应用自适应正交小波对纺织品纹理信息表征的研究

项目编号： No.61271006

项目类型： 面上项目

立项/批准年度： 2013

项目学科： 无线电电子学、电信技术

项目作者： 李立轻

作者单位： 东华大学

项目金额： 42万元

中文摘要： 纺织品纹理的数学描述是比较复杂的问题，却是织物表面检测、织物仿真等众多应用的先决条件和基础，小波变换应用于纺织品纹理的分析为其提供了从频域提取特征的方法。由于纺织品纹理的近似性和多样性，构造纹理自适应的正交小波基是一个有效的途径，其实质就是对小波基根据正交条件和设定的逼近条件进行优化。对于这一构造问题的求解，由于代价函数是非线性的，因而很难找到它的全局最小点，并且存在计算不稳定的问题，为此提出了分步满足正交条件和逼近条件的构造方法，即根据正交条件建立设定长度的正交小波基，然后根据织物纹理自适应的要求建立逼近条件，再在众多小波基中优选出满足逼近条件的小波基。以此为基础，建立基于计算机视觉的纺织品表面纹理自动检测实验设备，以验证自适应小波表征织物纹理的分析结果。该研究是以自适应小波分析同纺织品纹理分析相结合，对于织物疵点检测等机器视觉应用于纺织领域具有关键作用。

中文关键词： 纺织品；纹理；图像处理；自适应小波；表征

英文摘要： The mathematical description of textile texture is the base of fabric surface detection or fabric simulation although it is complex to analysis. Wavelet transform can be applied to solve the difficulty because it provides a method to analysis fabric texture in frequency domain. The construction of orthonormal wavelet which should be adaptive to the textile texture is a feasible way to solve the problem because textile texture is approximate and diversiform. The essential method is optimizing the wavelet base from its orthonormal conditions and approaching conditions. In the construction of wavelet bases it is difficult to find the global minimum of the cost function because it is nonlinear and the computation is instable. So firstly the orthonormal wavelet bases which length is determined are calculated according to the orthonormal conditions. Secondly the approaching conditions are built to adapt to the textile texture. And finally the optimal adaptive wavelet base is selected from the database. The inspection instrument for detecting textiles based on computer vision is based on the above theory and the test results are analyzed to testify the adaption to fabric texture. The research unites the wavelet transform and the textile texture analysis and is a key to the application of machine vision to the fabric in

英文关键词： textiles；texture；image processing；adaptive wavelet；characterization