Online three-dimensional recognition method of fabric defects

Abstract

The invention belongs to the field of image processing and pattern recognition, relating to a three-dimensional recognition method used for an online detection system of fabric defects. The method comprises the following steps: first converting three-dimensional data of fabrics to a coordinate system where the planes of the fabrics are positioned through coordinate transformation to realize computing of spatial dimension reduction, and then judging the types of the fabrics according to the distribution of X, Y and Z coordinates. Because of adopting three-dimensional coordinate information, the method can effectively get rid of the interference resulted from uneven wrinkles, willow catkins, ambient light or background light, not only can provide the information whether the defects exist but also can provide the types and reasons of the defects, thereby facilitating the operators to timely adjust production technology and avoiding the defects from appearing again.

Description

Technical field [0001] The present invention relates to an online three-dimensional recognition method of fabric defects. More specifically, the present invention relates to a three-dimensional recognition method used in an online detection system of fabric defects. Background technique [0002] China is currently a major exporter of textiles and garments, but we still have a long way to go from a textile and garment processing center to a profit center for textiles and garments. Since my country's accession to the WTO, textile exports have been competing with foreign products at low prices and low profits. The majority of them are middle and low-end markets, while high-end markets with strict quality requirements are difficult to enter. Therefore, it has become an urgent issue to improve the quality of my country's textiles by improving inspection technology and assembly. [0003] The detection methods of fabric defects mainly include statistical methods (statistical), model-bas...

AI Technical Summary

Problems solved by technology

[0008] (1) It is difficult to solve the image interference caused by "cloth surface wrinkles"

The cloth running (cloth rolling) roller is driven actively, and the cloth feeding is passive. Although there is a certain tension on the cloth surface in the detection area, the tension fluctuates during the cloth rolling process. If the tension is too large or too small, the cloth in the detection area will There are convex and concave wrinkles on the surface, and the wrinkles are imaged as shadows in the two-dimensional image, which are easily mistaken for defects

[0009] (2) It is difficult to solve the image interference caused by "changes in ambient light and background light"

When the ambient light or background light of the light box changes, or the background light has uneven light due to aging and other reasons, it will cause grayscale changes in the two-dimensional image. change distinguish

[0010] (3) It is difficult to solve the image interference caused by "flying catkins"

[0011] (4) At present, the above system is limited to the detection of defects in plain fabrics, and the detection of complex background colored fabrics needs further research

[0021] (1) The fabric data plane is relatively flat, but a certain piece of fabric data is missing, such as missing warp, missing weft, broken warp, broken weft, etc.

[0022] (2) The fabric data plane is relatively flat, but the fabric data is uneven, such as thin road, dense road, thick warp, thick weft, fine warp, fine weft, square eye, cloud weaving, etc.

Images