中文摘要：

在TFT-LCD平板表面缺陷检测中,一维DFT方法非常有效,空间域图像中周期Δx作为关键参数之一,需要准确地获取。传统的Δx的获取是通过人工手动计算得到的,为了自动获取Δx,设计了自动周期选取算法,利用数学统计对一维无缺陷图像进行处理,通过提取峰值位置、排序、计算间距和统计次数,得到准确的Δx。搭建了TFT-LCD平板检测实验系统,实验对不同长度（N=256、320和400）、不同分辨率（8.1、9.5和10.7μm/像素）、不同光照下的图像进行处理,能够准确地获取Δx,准确率达100%。将选取的Δx应用于缺陷检测中,系统可以准确检测出缺陷区域,证明了该选取方法具有良好的实用性和鲁棒性。

英文摘要：

One dimensional（ 1D） DFT method is very effective for the detection of surface defect on TFT-LCD panel. The period Δx in the spatial domain image,as one of the key parameters,needs to be acquired accurately. Traditional acquisition of the Δx is calculated artificially,in order to obtain the Δx automatically,an automatic selection algorithm of the period was designed,which utilized mathematical statistics to process 1D faultless image. By extracting the peak position,sorting,computing spacing and the number of statistics,the Δx is gotten accurately.A experimental system of TFT-LCD panel detection was built. The images with various length（ N = 256,320 and400）,resolution（ 8. 1,9. 5,and 10. 7 μm / pixel） and illumination were processed in the experiment. The experimental results show that the Δx is accurately obtained with an accuracy of 100%. By applying the selected Δx in defect detection,the system can detect the defect region accurately,so it indicates that the selection method has good practicability and robustness.