结合FFT和Gabor滤波器的织物纹理特征提取方法  

A Method of Extracting Fabric Texture Feature Combined with FFT and Gabor Filter

作　　者：任静[1] 周华[1] 郭超[1] 王妍[1] 梅再欢 

机构地区：[1]浙江理工大学先进纺织材料与制备技术教育部重点实验室,杭州310018

出　　处：《浙江理工大学学报（自然科学版）》2015年第1期6-10,共5页Journal of Zhejiang Sci-Tech University(Natural Sciences)

摘　　要：针对织物表面纹理特征提取与表征的问题,提出一种结合FFT(傅里叶变换)和Gabor滤波器的织物纹理特征提取方法,通过傅里叶变换提取频谱楔特征(角向分布)和环特征(径向分布)作为特征角度和频率参数,设计2个最优Gabor滤波器,分别与织物纹理图像进行卷积,获得目标子图像,然后对子图像进行融合处理,达到提取纹理特征的目的。在线条图像纹理特征成功提取的前提下,应用到常见织物的纹理特征提取上,并与传统的Gabor滤波器提取方法进行对比。实验结果表明,结合FFT和Gabor滤波器的织物纹理特征提取方法是有效的,且提取结果优于传统方法。Extraction method of fabric texture features combined with FFT and Gabor filter was proposed in allusion to fabric texture feature extraction and characterization. 2 optimal Gabor filters were designed through extracting frequency spectrum wedge feature （the angular distribution） and ring feature （radial distribution） with FFT as characteristic angle and frequency parameter. The 2 Gabor filters were respectively convoluted with texture image of fabrics to get the target sub-image, and then fusion process ing was made for the sub-image to achieve the purpose of extracting texture feature. On the premise of successfully getting the texture feature of line image, it was applied to extract common fabric texture fea- ture and compared with traditional Gabor filter extraction method. The results show that fabric texture feature method combined with FFT and Gabor filter is effective, and the extraction result is superior to traditional method.

关 键 词：傅里叶变换 GABOR滤波器 织物纹理 特征提取 

分 类 号：TS103.7[轻工技术与工程—纺织工程]