Intelligent detection method for positioning printed film based on machine vision

By performing grayscale equalization and feature extraction on images of the printing film production line, and combining them with a deviation discrimination model, the problem of single detection and evaluation criteria in existing technologies has been solved. Stable feature extraction and quantitative change detection of printing marks and film edges have been achieved, making it suitable for high-precision detection under complex working conditions.

CN122415591APending Publication Date: 2026-07-17YICHANG LANTIAN COLOR PRINTING CO LTD

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
YICHANG LANTIAN COLOR PRINTING CO LTD
Filing Date
2026-06-08
Publication Date
2026-07-17

AI Technical Summary

Technical Problem

Existing machine vision inspection methods lack standardized image enhancement processing in printed film production, and cannot simultaneously take into account the position and state of printed marks and the changes in the shape of the film edge. This results in a single basis for inspection and evaluation, and it cannot adapt to the stable extraction of image features and the synchronous quantitative determination of positioning and deformation states under complex working conditions.

Method used

By performing grayscale distribution histogram equalization on multiple frames of raw images continuously acquired on the printing film production line, the corner coordinates of the printing marks and the pixel trajectory of the film edge are extracted. The displacement vector and deformation curvature are calculated and input into a pre-trained deviation discrimination model to generate printing positioning deviation confidence and film deformation confidence, triggering a detection alarm signal.

Benefits of technology

It achieves stable feature extraction of printing marks and film edges, can quantitatively characterize position and morphological changes, expands the dimensions of detection and evaluation, and adapts to the high-precision, full-dimensional online detection needs in the continuous production process of printed films.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122415591A_ABST
    Figure CN122415591A_ABST
Patent Text Reader

Abstract

本发明涉及机器视觉检测技术领域,具体公开了基于机器视觉的印刷膜定位智能检测方法,包括采集印刷膜生产线连续多帧原始图像序列,对各帧图像进行灰度分布直方图均衡化处理,获取增强图像序列。从增强图像序列中分别提取印刷标记角点坐标集合与膜体边缘像素轨迹集合,依托角点坐标集合计算相邻帧印刷标记位移矢量,依托像素轨迹集合计算膜体边缘形变曲率。将位移矢量与形变曲率输入预训练偏差判别模型,输出印刷定位偏差置信度和膜体形变置信度,当任一置信度超出对应预设阈值,即刻触发检测报警并完成当前帧异常位置标注。该方法可改善工业光照干扰带来的图像质量缺陷,实现印刷标记定位偏移与膜体边缘形变的同步量化判别。
Need to check novelty before this filing date? Find Prior Art