基于织物结构特征和小波变换的快速疵点检测

Quick defect detection based on structure character of woven fabric image and wavelet transform

给出了一种利用小波为分析工具,基于织物采样图像进行疵点检测的新方法。首先对织物疵点图片进行了奇偶行列采样抽取,得到奇偶两幅子图像并使得图像的总数据变为原来的二分之一,然后再进行小波变换。在进行小波变换时,采用邻域插值,使得紧支撑小波滤波器的系数序列得到缩短,并可以由原来的滤波器系数计算出缩短后的滤波器系数。实验结果表明该方法可以在保证检测效果准确性的同时,大幅减少数据计算量、提升检测速度。

A new method of fabric defect detection using the structure character of fabric image and the wavelet transform is provided. The image is first sampled and extracted according to odd and even lines or columns. The image is decomposed into odd and even sub- images. This step reduces the total data of the image by half. Then, the two sub-images are analyzed by the wavelet transform. When being analyzed, the modulus serial of wavelet filters is shortened because of using the neighborhood interpolation, and the shortened modulus of filters can be calculated from the original modulus. The experimental results show that this method improves the speed of the defect detection while keeping the accuracy of the defect detection.