FFT和Hough变换在织物纹理方向检测上的应用

针对人工测量织物纹理方向的方法存在自动化程度低、精度不高的缺点，提出基于快速傅里叶变换和Hough变换的自动检测和识别织物经向、纬向、斜纹斜向的方法。对织物图像进行二值化处理，对二值化后的织物图像进行快速傅里叶变换得到功率谱图，对功率谱图的二值图像进行Hough变换，从而检测出织物纹理方向。通过对同一试样的不同放置方式进行测试，证明该方法对织物放置方式无特殊要求，仅需确保试样平摊在光滑台面上不受任何张力；将该方法与人工测试结果作对比，证明该方法具有高精度、高自动化的优点。

Application of FFT and Hough transform in fabric texture directions detecting

Considering the low automation and accuracy of manual measurement of fabric texture direction, fast Fourier transform and Hough transform are applied to automatically detecting and recognizing warp, weft and twill direction of a fabric. The binarization is used to obtain the binary image of a fabric. The power spectrum of the binary image is acquired based on Fast Fourier Transform（FFT）. The power spectrum is transformed to binary image and processed by Hough transform to get the fabric texture direction of the fabric. The experimental results show that there is no special requirement about fabric laying style and a specimen only needs to be flat on a smooth surface without any tension. It can be concluded that the method has the advantages of high precision and high automation compared with manual test results.