Printed image inspection method with defect classification

Patent metadata

Field	Value
Publication number	US-11580630-B2
Application number	US-202117167379-A
Country	US
Kind code	B2
Filing date	Feb 4, 2021
Priority date	Mar 11, 2020
Publication date	Feb 14, 2023
Grant date	Feb 14, 2023

Abstract

A method of inspecting images on printed products by a computer in a printing machine. Printed products are recorded and digitized by an image sensor of an image inspection system in the course of the image inspection process, and the computer compares them to a digital reference image. If deviations are found, the defective printed products are removed. The computer analyzes the deviations found in the course of the image inspection process together with further data from other system parts and from the machine, determines specific defect classes and the causes thereof based on the defects by machine learning processes, assigns the defects found in the image inspection process to the defect classes in a corresponding way, and displays the classified detected defects with their defect classes and causes to an operator of the machine so that the operator can initiate specific measures to eliminate the defect causes.

First claim

The invention claimed is:

1. A method of inspecting images on printed products by a computer in a machine for processing printing substrates, the method which comprises:
recording and digitizing printed products with at least one image sensor of an image inspection system to generate recorded printed images in an image inspection process;
comparing the recorded printed images with a digital reference image by the computer and, if deviations are found between the recorded printed images and the digital reference image, removing the printed products that have been found to be defective; and
causing the computer to analyze the deviations found in the course of the image inspection process and further data from other system parts and from the machine as detected defects, to determine specific defect classes and defect causes thereof based on the detected defects by machine learning processes, to assign the defects found in the image inspection process to the defect classes in a corresponding way, and to display the classified detected defects with the respective defect classes and causes to an operator of the machine to enable the operator to initiate specific measures to eliminate the defect causes.

2. The method according to claim 1 , which comprises causing the computer to superimpose the defects that have been classified as a group onto the digital reference image and display the defects superimposed on the reference image to the operator of the machine on a display.

3. The method according to claim 2 , which comprises providing for every defect class an icon or key word on the display to disclose the defect class to the operator and displaying the group of individual classified defects to the operator in combination with the respective icon or key word.

4. The method according to claim 2 , which comprises displaying with the computer the detected classified defects as a group with a local reference in the digital reference image.

5. The method according to claim 1 , wherein the defect classes comprise typical problems inherent in a printing process.

6. The method according to claim 5 , wherein the typical problems are selected from the group consisting of foreign objects, smearing, bent paper edges, register measurement, color measurement, monitoring of defective nozzles in a digital printing machine, and white lines in the printed image.

7. The method according to claim 1 , which comprises causing the computer to derive from the specific defect class and from the determined cause a suggestion for a reaction and to display the suggestion to the operator on a display, whereupon the operator implements the suggestion after manual assessment.

8. The method according to claim 1 , which comprises causing the computer to derive a suggestion for a reaction from the specific defect class and from the determined cause and to automatically put the reaction into practice.

9. The method according to claim 1 , which comprises causing the computer to record data on the classified defects, to statistically analyze the data, to derive suggestions on how to avoid defects from the data, and to display the suggestions to the operator on a display for the operator to implement or dismiss the suggestions after manual assessment.

10. The method according to claim 1 , wherein the classification of the defects by the computer is dependent on parameters that are initially defined by default factory values when the machine is delivered and are subsequently trained by the computer in the course of the image inspection process.

11. The method according to claim 10 , wherein the parameters are trained by the computer by changing the presettings on the machine, by adapting the parameters via print job data, by accessing a central database, or by interacting with the operator.

Assignees

• Heidelberger Druckmasch Ag

Inventors

• Soltwedel Frank
• Mueller Robert
• Krieger Jan
• Schumann Frank
• Eisele Peter

Classifications

• B41F33/0036 Primary

  Devices for scanning or checking the printed matter for quality control · CPC title

• B41F33/02

  Arrangements of indicating devices, e.g. counters · CPC title

• G01N21/8983

  for testing textile webs, i.e. woven material · CPC title

• B41J2/2142

  Detection of malfunctioning nozzles (generating single droplets or particles on demand by pressure, e.g. electromechanical transducers B41J2/045, B41J2/05; jet deflection sensors B41J2/125; for cleaning purposes B41J2/16579) · CPC title

• G06T7/001 Primary

  using an image reference approach · CPC title