基于快速傅里叶变换与织物纹理能量算法的起毛起球织物纹理滤除

在织物起毛起球图像处理过程中，织物纹理信息的滤除是实现毛球分割至关重要的一步，而不同的图像包含的纹理、毛球信息不同，在用傅里叶变换对图像进行处理时需要采用不同滤波半径。因此，在处理过程中选取的滤波半径是实现纹理滤除的关键因素。本文将快速傅里叶变换与织物纹理能量的算法相结合，根据能量2次差值拟合曲线实现毛球与纹理分离时最佳滤波半径的自动获取，然后通过反傅里叶变换得到分离的纹理信息和毛球信息，实现织物纹理的滤除，最后分割出完整的毛球信息。

Pilling Fabric Texture Hlter Based on Fast Fourier Transform and Fabric Texture Energy Algorithm

In the process of pilling fabric image processing, the filter of fabric texture information is very important to realize pilling segmentation, and different image contains different texture and pilling information, then needs different filter radius in the image processing. Therefore, the selection of filter radius is the key factor to realize the texture filter in the image processing. In this article, the fast fourier transform and the fabric texture energy algorithm were combined, realizing the automatic acquisition of best filter radius according to the energy fitting curve, in the process of pilling and texture separation. And then with inverse fourier transform the researchers got the separated texture and pilling information, realizing the filter of fabric texture, segmenting the complete pilling information finaUy.