Research On Fabric Defect Recognition Based On The Conformal Monogenic Feature

Fabric defects are important factors affecting the quality and price of fabric products.At present,most of the fabric defects are identified by artificial identification,but there are some shortcomings such as low efficiency,limited precision and lack of objectivity due to subjective factors.Although experts and scholars at home and abroad have done a lot of research on fabric defect identification,there are still problems such as less identification and limited precision.Therefore,the topic selection has important practical significance and application value.On the basis of reading a lot of literature,Firstly,the research situation of fabric defect feature extraction method and fabric defect identification method is researched and analyzed at home and abroad,and then the theoretical research and experimental research on fabric defect automatic recognition based on conformal feature is carried out.The main research work is as follows:(1)The edge information of the fabric defect image reflects the weaving rule of the yarn at the defect point.In order to overcome the influence of the uneven illumination or the contrast ratio on the edge detection result,according to the good local characterization ability of the conformal signal.Based on the improved phase consistency model,a conformal phase congruency model is proposed.The experimental results show that compared with the existing phase-consistent edge detection method,this method not only improves the accuracy of edge detection,but also makes the edge more uniform and continuous,has a certain anti-noise ability,and is not affected by the change of light.Based on the good performance of conformal phase uniformity,the local binary pattern is used to encode the conformal monogenic phase consistency binary coding feature,which can show the interlaced law of the edge of different defects and the defect area,Thus describing the differences between the different defects.(2)In order to accurately locate the fabric defects and obtain the shape features of the fabric defects,a new corner detection method based on chord-to-point distance accumulation technology is proposed.Using the distribution of the corner points on the fabric defect map to accurately locate the fabric defects and obtain the aspect ratio of the fabric defects.This method using the conformal monogenic phase congruencyinstead of Canny to complete the image edge contour extraction,eliminate the impact of edge positioning inaccurate on subsequent corner location;Through the candidate corner neighborhood re-detection technology to avoid the adjacent corner of the missing test;introducing the adaptive threshold effectively removes the oval corner and preserves the blunt corners;the method of Freeman chain code is used to remove pseudo corner points.Through the subjective and objective analysis,this method has a better corner detection effect.(3)In order to better characterize the texture of fabric defect image,an image texture feature extraction method based on conformal feature fusion is proposed.The local amplitude,local phase,local direction and local curvature are obtained by using the conformal signal.The local characterization of the four local features is carried out according to the good local characterization ability of the conformal signal.Coding feature,the fusion of four local binary coding features to form a conformal fusion feature to characterize the texture features of the image,the texture features can better describe the different fabric defects of the texture differences.(4)Connected conformal monogenic phase consistency binary coding feature and conformal monogenic fusion feature to characterize the texture features of the image.Through the experiments on the Brodatz texture library,the validity of the feature connection is verified,and then the expression is connected The results show that the recognition method of this paper is effective and the recognition of 12 kinds of fabric defects is realized.The recognition rate of the fabric defect is high,and the recognition rate of the fabric defect is high 99.41%.Figure 37,Table 17,reference 94...