Detecting method and detecting device for printed image

Abstract

The invention discloses a detecting method and a detecting device for a printed image, wherein the detecting method comprises the following steps of acquiring the printed image; performing coarse extracting on the edge contour of a foreground area according to an automatic threshold segmentation method; editing the edge contour according to a gray scale difference threshold, thereby accurately extracting the edge contour line of the foreground area; positioning the edge contour line according to an area positioning method and a linear positioning method; and detecting whether a defect exists in the edge contour line according to a Blob analysis method. The detecting method provided by an embodiment of the invention improves extraction accuracy and positioning accuracy of the edge contour, thereby realizing accurate detection for edge defects.

Description

technical field [0001] The invention relates to the field of image technology, in particular to a detection method and a detection device of a printed image. Background technique [0002] At present, in the printing inspection industry, edge defects are a relatively common defect type. Incomplete die-cutting, the influence of printing machines, and the carelessness of operators will all cause uneven edges of printed products, resulting in rough edges, missing edges, and missing edges. Defects such as corners, wrinkles, scratches, dirty edges, etc. Since such defects are distributed at edge positions, they are collectively referred to as edge defects. Edge defects are characterized by their proximity to the edge and their small size with low contrast to the product background. Higher requirements are put forward for the traditional detection method: first, the distance from the edge means that a fine positioning algorithm with higher positioning accuracy is required, which ...

AI Technical Summary

Problems solved by technology

[0002] At present, in the printing inspection industry, edge defects are a relatively common defect type. Incomplete die-cutting, the influence of printing machines, and the carelessness of operators will all cause uneven edges of printed products, resulting in rough edges, missing edges, and missing edges. Defects such as corners, wrinkles, scratches, dirty edges, etc.

However, there is no obvious feature area in the flat area, which makes positioning difficult, resulting in a certain positioning deviation, making the edge of the learned image blurred, and the detection effect on edge defects is poor.

Although the effect is improved after using the linear positioning method to locate the edge contour, the detection algorithm based on learning comparison still cannot detect edge defects well due to the small size of edge defects and low contrast

At the same time, in the traditional learning comparison detection method, after the detection area is drawn as the product boundary range, due to the influence of the belt area and the belt hole area, using the learned image as the good product range will result in more false detections

Images