Defect detection system and method for thin film material

Abstract

According to the defect detection system and method for the thin film material, a conveying line used for conveying the thin film is installed between a linear array camera and a light source according to the characteristics of the thin film, and the light source emits high-brightness linear light from bottom to top, so that the linear array camera clearly photographs the thin film material in a backlight environment; and the photographed image is gray, so that the texture of the film can be highlighted, and the contrast ratio of the defect to the background is increased. In addition, a deep learning method is used, and an online data enhancement method is designed, so that the data volume of small samples is secondarily increased, the defect detection effect is improved, and the probability of missing detection and false detection is reduced. And finally, performing defect detection on the thin film material by using an improved Yov4 detection model, so that the defect detection effect is effectively improved.

Description

technical field [0001] The invention relates to the technical field of machine vision detection and image processing, in particular to a defect detection system and method for thin film materials. Background technique [0002] Thin-film materials (also known as separators) are one of the important components of lithium batteries and account for a large proportion of battery manufacturing costs. This material is a layer of porous plastic film with good insulation and good alkali resistance in the electrolyte. It can isolate the aluminum and nickel electrodes in the battery and provide a flow path for electrons. Batteries are widely used, such as common electronic products (mobile phones, notebooks, power banks) and new energy vehicles. Driven by market demand, the production of thin films has gradually increased. The performance of the film affects the performance of the lithium battery, which is related to the manufacture of lithium batteries and the safety of the product....

AI Technical Summary

Problems solved by technology

[0005] 1) The feature extractor mainly relies on manual design, which largely depends on the technical knowledge of technicians and the complex parameter adjustment process, and the design scheme is aimed at specific industrial applications, and its generalization ability is poor

[0006] 2) In the detection of film defects, for defects such as creases, the gray level is very similar to the background, and the detection effect of traditional image algorithms is not good

[0007] 3) When there are few defect samples or data sets, the detection accuracy is low

Images