基于互相关的印花织物疵点检测

为实现印花织物中疵点的自动检测，以互相关理论为基础，结合图像处理技术，以Matlab7.0构建了一套印花织物疵点自动检测系统。在疵点检测过程中，提出以加和表理论为基础实现互相关系数的快速计算。通过对软件模拟的印花花纹疵点的识别，说明这个系统能够实现印花过程中常见的花纹偏移、颜色色差等疵点的自动检测。实际印花织物疵点的检测实验表明，所提出的算法具有有效性、鲁棒性等优点。通过比较不同子窗口大小的检测结果，选定25像素×25像素作为最终检测系统中子窗口的大小。

Defect detection of printed fabrics using normalized cross correlation

A novel detection system based on normalized cross correlation and image analysis is proposed to detect the defects of printed fabrics. Matlab 7.0 is used as the software tool to construct the system. During the detection process, a fast normalized cross correlation computation is proposed based on the theory of sum-table scheme. The detection of the printed fabric defect simulated in software proves that the system can inspect the normal defects, such as color displacement, color aberration, of printed fabrics automatically. The experimental detection for actual printed fabric shows that the effectiveness and robustness of the proposed in this paper. By comparing the detection results of sub-windows with different sizes, 25 pixels×25 pixels are selected as the parameter of the sub-window in the final defect detection system.