Method and system for detecting printing defects

By employing adaptive dynamic region segmentation, image optimization preprocessing, and dual verification using a deep learning model, the accuracy and speed issues of traditional printing defect detection methods have been resolved, enabling accurate detection and high-speed adaptation of printing defects across all product categories.

CN122415429APending Publication Date: 2026-07-17SHENZHEN AWELL INTELLIGENT TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
SHENZHEN AWELL INTELLIGENT TECHNOLOGY CO LTD
Filing Date
2026-03-09
Publication Date
2026-07-17

AI Technical Summary

Technical Problem

Traditional printing defect detection methods struggle to balance full-category defect coverage, detection accuracy, and detection speed. They are also highly sample-dependent and have high false positive and false negative rates, failing to meet the precise detection requirements of high-speed printing production lines.

Method used

By employing an adaptive dynamic region segmentation strategy, image quality optimization preprocessing, and a dual verification method that integrates deep learning models and traditional visual rules, we can achieve rapid identification, classification, and graded detection of printing defects.

Benefits of technology

It achieves accurate detection of printing defects across all product categories, is compatible with high-speed printing production lines, improves detection speed, reduces reliance on samples, and can quickly adapt to different printing processes.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122415429A_ABST
    Figure CN122415429A_ABST
Patent Text Reader

Abstract

The present application belongs to the technical field of image processing, and particularly relates to a printing defect detection method and system. The present application collects a to-be-detected printed matter image, performs an image quality optimization preprocessing operation on the to-be-detected printed matter image, and outputs a standardized detection image; based on an adaptive dynamic region division strategy, the detection image is subjected to rapid rough detection, and a non-defect region and a suspected defect region are screened out; the suspected defect region is input into a fusion deep learning model, identification, classification and positioning of all categories of printing defects are realized, and a fine detection result is output; based on double checking of the fine detection result by traditional visual rules and deep learning features, the defects are classified and graded, and a final detection report is output.
Need to check novelty before this filing date? Find Prior Art