| The Chinese economic now transform from rapid growth to stability. With this change, the competition of textile production becomes more and more intense. Many textile factories face two big problems: reducing production cost and improving the quality of products. At present, in domestic market, some factories start using auto textile flaw detection equipment. Equipment from foreign countries has good detection effect, but they are too expensive for most of the factories to afford. Domestic equipment is cheaper, but the effect is limited, it can’t fit the demand of widespread use.For the textile flaw detection of warp knitting machines, we propose a flaw detection method based on image correction. First step, we record the distortion of the template, by this, we can acquire the stable distortion information of a camera. Second step, with the distortion information of the camera, we correct the distorted image captured by the same camera. If the image have boundary of textile, we also correct the boundary with similar method. Third, we detect the image. After the vertical projection of the image, we deal with the one dimension data by the median filter algorithm. We can get a difference value array by using the data after median filter algorithm minus the original one dimension data. We deal with the difference value array by Gaussian filter. Using this data, we decide whether the original image have a flaw or not.Based on the warp knitted textile flaw detection algorithm, we developed the textile flaw detect program. The program is developed based on MFC with VC6.0, it can be used in Windows XP/7 OS. It can be installed on industrial computer, applied to many kinds of warp knitting equipment. Different factories from different places have already used this system, effect shows that this system can detect the flaw in a short time, nearly without any mistake. This system can be used in warp knitted factories to detect textile flaw automatically. |
PDF Full Text Request