Automatic measurement of yarn hairiness based on the improved MRMRF segmentation algorithm

In order to determine the yarn hairiness characteristic index more accurately, a new yarn hairiness testing strategy based on a combination of image acquisition technology is proposed. Firstly, the captured yarn images are processed with gray-scale conversion and skew correction. Secondly, yarn segmentation is implemented using a multi-resolution Markov Random Field (MRMRF) model with a variable weight in the wavelet domain and yarn stem separation is realized through iterative threshold segmentation algorithm. Thirdly, the image of yarn hairiness is extracted. Finally, the total number and actual number of yarn hairiness of different length are counted sequentially based on the segmentation lines and baseline of yarn stem edge. The baseline is obtained by calculating the average distance between yarn stem edge and yarn axis. Furthermore, the feature is analyzed. Experimental results show that, compared with visual observation method, the maximum deviation of proposed image processing algorithm is 3.88%. The proposed approach can make the results of yarn hairiness segmentation more precisely, and then the more reliable results of hairiness feature detection are obtained.