Self-learning chemical fiber spindle intelligent appearance detection system and method based on big data

Abstract

The invention relates to an automatic classifying type spindle appearance detection system and method, which solve the problems of waste of labor, material and cost, lack of position and light source control, influence to subsequent appearance detection result, and low efficiency and accuracy on appearance defect detection of the spindle during sorting of spindles with appearance defects in a manual manner. The system comprises a tray and a conveying unit, wherein a dark box is arranged on the conveying unit; a sorting unit is arranged on the conveying unit behind the dark box; an image collection unit is arranged in the dark box; the appearance defect of the spindle is analyzed by collecting an image of each surface of the spindle, and the spindle is automatically classified according to the analysis result. Compared with the manual manner, the system has the advantages that a large amount of labor, material and cost is reduced; compared with common check system, the function of automatically classifying the defective spindles is added, the workload is reduced, and the working efficiency is improved; by providing a self-learning function, the accuracy of oil dirt detection is further improved.

Description

technical field

[0001] The invention relates to a wire ingot appearance defect detection technology, in particular to an automatic classification type wire ingot appearance detection system and method. Background technique

[0002] Defects on the chemical fiber spindles not only affect the appearance of the spindles, but also affect the grade of the spindles, and even broken filaments may occur, affecting the production of downstream links. At present, the silk ingots with appearance defects are generally classified manually, but using manual methods is undoubtedly a huge waste of manpower, material resources and financial resources. If you want to use highly automated detection methods, you need to use sophisticated control systems, efficient detection algorithms and accurate classification techniques.

[0003] The existing silk ingot detection methods only detect a single variety of silk ingots under specific circumstances, and the detection results cannot be automaticall...

Images