基于频域滤波的织物疵点检测

根据织物纹理和疵点的频谱不同分布，提出了织物疵点检测的新方法。首先在研究织物纹理分类的基础上，通过傅立叶变换获得频谱图像；然后设计频域滤波器抑制正常纹理频谱信息，通过重构灰度图像，把灰度图像分割成若干子窗口；最后将统计子窗口方差作为特征值与设定的阈值比较，从而判别是否存在疵点。实验结果表明该方法是有效的, 检测正确率达到90%以上。

Fabric defects detection based on frequency domain filtering

According to the spectrum of fabric texture and defects, a new method for defect detection was presented. Firstly, on the basis of the fabric texture classification, a spectrum image was derived from the Fast Fourier Transform (FFT). Then, the frequency filter was designed to remove normal texture information, and the acquired reconstruction images were segmented into many sub-windows, in which variances were calculated as characteristics. Lastly, the characteristics were compared with the standard threshol...