Method for off-line automatic detection of filament coil tail filament

Abstract

The invention relates to a method for off-line automatic detection of a filament coil tail filament. The method includes dividing the surface of the coil into a plurality of areas; collecting an appearance image of each area; respectively inputting the classification models into the classification models corresponding to the regions; outputting a label representing whether a tail silk defect exists or not by the classification model, wherein the plurality of regions are n regions; i=1, 2, ..., n, the classification model corresponding to the region i is a LeNet convolutional neural network trained by a database, the database comprises a plurality of appearance images and labels corresponding to the appearance images, the appearance images are appearance images of a plurality of silk roll surface areas i, the appearance images and the labels are used as an input item and a target output item respectively during training, and the coils are a set of coils without tail coil defects in thesurface areas i and coils with tail coil defects in the surface areas i. The method has the advantages of high detection efficiency, high accuracy and excellent popularization value.

Description

technical field [0001] The invention belongs to the technical field of chemical fiber detection, and relates to an off-line automatic detection method for coiled tails of filaments. Background technique [0002] Tail silk is an independently visible single or multiple filaments that appear on the top and tail of the filament roll and are separated from the main body of the roll. It is a common defect in the filament roll and will greatly affect The appearance and quality of the filament coil, for example, in the subsequent process, the production of the product is often interrupted due to the existence of the tail thread, so the filament coil must be inspected for the tail thread of the package appearance before packaging. [0003] At present, the detection of tail silk is mostly carried out by human eyes. It is necessary to manually check each silk roll package to determine whether the silk roll has tail silk. This will not only consume a lot of manpower and time, but also ...

AI Technical Summary

Problems solved by technology

[0003] At present, the detection of tail silk is mostly carried out by human eyes. It is necessary to manually check each silk roll package to determine whether the silk roll has tail silk. This will not only consume a lot of manpower and time, but also affect production efficiency. The fatigue of human eyes and human subjectivity will also affect the accuracy of tail wire detection to a certain extent

Images