基于多分辨率分析及灰度共生矩阵的织物纹理识别.pdfVIP

th Proceedings of the 25 Chinese Control Conference 7-11 August, 2006, Harbin, Heilongjiang 基于多分辨率分析及灰度共生矩阵的织物纹理识别 苏慧 费树岷 东南大学自动控制系 江苏南京 210096 E-mail: suhui52@ 摘要：本文提出了一种将织物图像从宏观上看作一种纹理图像，提取其纹理特征对不同织物进行区分的方法。 在进行纹理特征提取时，文中在传统的基于灰度共生矩阵的纹理特征提取的基础上加入了多分辨率分析的思 想。根据不同分辨率下纹理图像会呈现出不同的特征的原理，首先对图像进行基于小波的多分辨率分解，然后 再对分解后的每级图像分别求灰度共生矩阵，并提取相应特征，以此为基础进行图像的识别。实验结果表明本 文所提算法具有较高的识别准确率。 关键词：纹理识别，灰度共生矩阵，多分辨率分析，织物图像 The Texture Recognition of Fabric Based on Multi-resolution Analysis and Gray-level Co-occurrence SU Hui FEI Shu-min Department of Automatic Control, Southeast University, Nanjing 210096, China E-mail: suhui52@ Abstract: In this paper, we propose a method that can classify a variety of fabrics. In this method, the fabric images are regarded as texture images, and distinguished by different texture characters. When extracting the feature, we add the idea of multi-resolution analysis to the ordinary gray-level co-occurrence method. The basis of the idea is that textured images present different features at different resolutions. The images are processed first with multi-resolution analysis, and then with gray-level co-occurrence. After that, the image’s feature is extracted. Experiment results show that the texture images can be recognized more correctly with this new method. Key Words: texture recognition, gray-level co-occurrence, multi-resolution analysis, fabric images 1． 引言(Introduction) 由于织物种类繁多,每种织物的经纬线及其缝隙在 图像上呈现的差异较大,加之纱线常发生扭曲,变形 近几年由于数字图像处理技术及设备的迅速