Abstract:

In order to detect the broken yarn visually in the production of the spinning, according to the particularity of the yarn image in the direction, a method for realizing the yarn breakage detection was developed. Industrial cameras were used to capture the monement of the yarn, firstly wavelet denoising was used for smoothing image,  then the exact information of the yarn were extracted by the simplified. Hough transform and collinearity test,  and finally according to the actual yarn distance characteristics, it was determined whether the yarn break. The sym 4 wavelet bases ware chosen from many wavelet transforms, the results show that according denoising effect is optimal when the threshold is 10, and the edge detection operator prewwit is used to detect the vertical direction, optimize the Hough transform to reduce the detection angle to [-10 °, 10 °], enlarge the angular interval to 4 °, and shorten the operation time from 0.46s to 0.31s, which reduces the operation time and improves the operation speed. Experiment resolts show that the algorithm can accurately determine the yarn breakage information.

Key words: yarn breakage detection, image processing, Hough transform, wavelet denoising, edge detection operator