A method and device for detecting concave and convex point defects of packaging box

CN114066810BActive Publication Date: 2024-11-29HENGYINDA (GUANGDONG) PRINTING TECH CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202111182335.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-10-11
Publication Date
2024-11-29
Estimated Expiration
2041-10-11

AI Technical Summary

Technical Problem

The prior art is difficult to effectively detect complex and multi-scale concave and bump defects in tablet packaging boxes, resulting in high manual detection costs and high error rates, making it difficult to apply to existing computer vision detection technologies.

Method used

By constructing a high-precision concave and bump detection model, a lightweight deep neural network is designed using the FCNN framework, combining high-pass filtering and Wiener filtering to process images, feature images are extracted and closed operations are performed, regional feature images of the packaging box are obtained, and samples are collected for Training to achieve rapid detection of complex and multi-scale concave and bump points.

Benefits of technology

High-precision detection of complex and multi-scale concave and bump defects in tablet packaging boxes is realized, reducing detection costs, improving detection efficiency, and highly scalable transfer learning for other types of problems.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114066810B_ABST
    Figure CN114066810B_ABST
Patent Text Reader

Abstract

The present invention discloses a method and device for detecting concave-convex point defects of a packaging box. By collecting packaging box samples with concave-convex point defects, image compression, feature fusion, image segmentation and other processing methods are performed on the samples to obtain a set of regional feature images. A lightweight deep neural network is designed based on an FCNN framework, a detection model is constructed, and the model is trained using the regional feature images of the samples to obtain a high-precision concave-convex point defect detection model. The packaging box is detected according to the high-precision detection model, thereby realizing rapid detection of complex and multi-scale concave-convex point defects in the packaging box.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Express outer package defect detection method and device based on deep learning

    CN111862092A

  • Figure interaction detection method based on deep learning

    CN111914622A