基于小波分析与纹理能量变换的织物疵点检测

Fabric Defect Detection Based on Wavelet Analysis and Texture Energy Transform

为了准确检测织物疵点,将含疵点织物图像进行二维小波分解,在小波分解后的经向和纬向子图上将图像分成大小相同的矩形局部重叠窗口,对矩形窗口进行laws纹理能量变换,并将变换结果与给定的阈值进行比较,进而检测和识别出疵点。试验证明,该方法对素色织物的断经、缺纬具有快速、准确的检测效果,也可以检测双经、双纬等疵点。

To detct fabric defects exactly, defect fabric image was decomposed into two-dimensional wavelet, after decomposing, divide image of weft and warp into rectangular sectional window, and texture energy transform was done to these window. Comparing changed result to setting threshold, and then detect and identify defects. The detection shows that this method has quick and exact test effect to broken warp ＆ lack wefts of solid colored cloth, and it can also detect defects of double warp, double wefts and so on.