Geometric correction method and system for detecting lateral deviation of image of printed matter

Abstract

The invention discloses a geometric correction method and system for detecting the lateral deviation of an image of a printed matter, and the method comprises the steps: building a smoothing kernel for row pixels of an image collected by a linear array camera according to the vibration amplitude, causing the distortion of the lateral deviation, of printing equipment; carrying out the lateral smoothing of the pixels of a reference image and a to-be-detected image row by row through the smoothing kernel; respectively calculating the absolute value pairs for the lateral smoothing of each row; selecting a minimum subtraction value, corresponding to each row, from the absolute value pairs of each row; determining the lateral coordinates of the pixel positions according to the lateral coordinates of the pixel positions of each minimum subtraction value during the obtaining of the minimum subtraction values; calculating the offset of the row pixels according to the lateral coordinates of thepixel positions so as to correct the lateral distortion of the pixels of the to-be-detected image to the reference image, thereby facilitating the subsequent detection of the defects of the printed matter through the to-be-detected image, guaranteeing the comparison premise of a single variable of two images, reducing the differential artifact greatly, and reducing the defect detection error rate.

Description

technical field [0001] The invention relates to the technical field of printed matter detection image detection, in particular to a geometric correction method and system for lateral offset of printed matter detection images. Background technique [0002] In the print image defect detection module, the inconsistency between the product to be inspected and the qualified product is found by comparing the image of the qualified product (also called the reference image) with the image of the product to be inspected (also called the image to be inspected), which is defined as a defect. During the shooting process of the printed product image, the actual shooting environment of the qualified product and the product to be inspected is not absolutely consistent, and distortion such as lateral distortion may occur. Therefore, the printed product to be inspected must be back-corrected to the qualified product image before defect detection, so The images of qualified products and untes...

AI Technical Summary

Problems solved by technology

However, compared with the exposure frequency, the mechanical vibration frequency is relatively small, so the continuous exposure of the linear array CCD sensor 13 within an offset time period causes an image to be skewed, and a pair of distorted printed matter images can be obtained after multiple accumulations

[0006] In the subsequent defect detection, the degree of distortion and distortion of the reference image and the image to be inspected are inconsistent, which will lead to non-unique comparison variables between the two, which will cause great difficulties for real defect detection

Images