采用频域滤波的织物疵点检测方法

首先,通过傅里叶变换获得含有疵点的织物频谱图,根据频谱图设计频域滤波器分离正常的纹理信息,保留疵点信息;然后,通过傅里叶逆变换进行灰度图重构,得到去除正常纹理的疵点图像;最后,经过高斯平滑后二值化分割出织物疵点.实验选取5种织物疵点进行验证,结果表明:采用频域滤波的织物疵点检测方法检测效果良好,且检测率达到90%以上.

Detection Method of Fabric Defects Based on Frequency Domain Filtering

Firstly, a spectrum image for fabric containing defects was obtained from the Fourier transform, According to the spectrum image, frequency domain filter was designed to remove normal texture information and keep defect information; and then a fabric defect image is obtained by Fourier inverse transform for reconstructed gray image which remove normal texture, finally, the fabric defect was obtained after Gaussian smoothing and image segment. The experiment select five types of fabric defects, the experimental results shows that the method is effective, and the detection average accuracy is over 90%.