Abstract:

Aiming at the problem that the existing yarn hairiness detection methods cannot effectively detect curving and crossing hairiness, this paper presents a yarn hairiness path tracing algorithm based on an image analysis method. The backbone processed images were obtained by the pre-processing followed by backbone processing. The hairiness endpoint was taken as the starting point, and the new hairiness path point was obtained by judging the eight neighboring pixels of hairiness starting point, the neighborhood of hairiness path points was judged repeatedly until no hairiness path points existed. In the case of multi-path intersection of hairiness, the cross-matching value index was proposed. According to the slope of the adjacent hairiness path points in front of the cross point of hairiness and assigning dynamic weight to get the local slope of hairiness, cross-matching value was used to match the multi-path hairiness points to get the new hairiness points, and the number of hairiness pixels was obtained and converted into the hairiness length by the hairiness tracking method. According to the comparison of the detection results coming from the manual method and projection method, the error between the detected result of hairiness tracking and the manual inspection was less than 4%. This result indicated an effective solution to the problem in tracking and detecting the crossing and curving hairiness, improving the detection accuracy of yarn hairiness.

Key words: yarn hairiness, path matching, eight neighborhoods, dynamic weight, hairiness slope, hairiness detection, image analysis method