Unlock instant, AI-driven research and patent intelligence for your innovation.

Banknote quality inspection system and method based on deformation recovery technology

A detection system and technology technology, applied in the direction of instruments, calculations, characters and pattern recognition, etc., can solve the problems of difficult to distinguish between different printings, difficult positioning and alignment of offset printing areas and gravure printing areas, and achieve the effect of tolerance to geometric distortion

Active Publication Date: 2019-06-25
XIAN BANKNOTE PRINTING +1
View PDF6 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] In order to solve the technical problems of the existing image defect detection method that it is difficult to distinguish between different printing times during the positioning process, and the offset printing area and the gravure printing area are difficult to locate and align, the present invention provides a banknote quality detection system based on deformation restoration technology and method

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Banknote quality inspection system and method based on deformation recovery technology
  • Banknote quality inspection system and method based on deformation recovery technology
  • Banknote quality inspection system and method based on deformation recovery technology

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0072] The first step of detection is positioning: that is, each detection area is determined by the positioning kernel. This step is basically the same as the current positioning method, but sub-pixel alignment is not required. The computational complexity is related to the search range, the size of the positioning core, and the number of positioning cores.

[0073] The second step of detection is reconstruction: for the convenience of illustration, the following takes a 100×100 image block as an example.

[0074] 1) Data reconstruction first needs to project the data x into the trained linear subspace data.

[0075] y=U k '(x-μ)

[0076] where U k ' Get Uk for the training phase.

[0077] 2) Reconstruct to the original space using the first k eigenvectors.

[0078]

[0079] in to reconstruct the image.

[0080] Specifically:

[0081] It is necessary to remove the mean value of the real-time samples, that is, perform 10,000 subtractions. (An image of 100×100 is ...

Embodiment 2

[0085] next to Figure 6 The R component of the region is detected, and there are 800 training samples in total. The location kernel 1 size is 100×100, and the image size is 600×1100. Positioning core 2 size 80×80

[0086] In this embodiment, two positioning kernels are used, such as Figure 7 a and Figure 7 as shown in b.

[0087] For two positioning nuclei, there are two measurements, where Figure 8 a-8d is the process of reconstructing an image with better quality and no defects, and finding the difference to realize the detection; Figure 9 a-9d is the process of reconstructing an image with obvious quality defects and finding the difference. Here the two results are for different test images. Just the difference between the two test samples.

[0088] In one inspection, more than one positioning core is often needed to complete the positioning. In the RMB inspection, there are two printings of offset printing and gravure printing, and there is an overprinting rel...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention relates to a banknote quality inspection system and method based on deformation restoration technology. The algorithm used in banknote quality inspection removes the irrelevance of data in the training set by retaining the main features of the training set image data and common changes. The real-time image is reconstructed, the relevant data part of the image in the training set is retained, and the irrelevant part of the image is removed. The real-time image is compared with the reconstructed image point by point to obtain a difference image, and the defect is detected after thresholding. The present invention solves the problem of inaccurate layering of glue and gravure in the positioning process of the existing method, realizes that there is no need to distinguish glue and gravure in the detection process, geometric distortion can be tolerated, and the boundaries of glue and gravure can be directly checked. At the same time, it belongs to the location near the glue and gravure printing area, which is difficult to locate and detect. The detection accuracy is improved, false detection is reduced, the machine can be prompted to quickly and accurately find problems, and the product quality can be controlled in a timely and effective manner.

Description

technical field [0001] The invention relates to a banknote quality detection system and method based on deformation recovery technology. Background technique [0002] Currently, in the existing image defect (difference) detection process, it is necessary to accurately register the real-time image with the reference image and perform point-to-point comparison. When the point-by-point comparison difference is greater than a certain threshold, it is considered that there is a difference in the corresponding position of the real-time image. Defect detection can be performed after a series of processing on the difference image. The key to this detection method lies in the precise registration of the image. The detection can only be performed after the real-time image is completely aligned with the reference image (sub-pixel level). There are the following difficulties in the alignment process: [0003] 1. Due to the different printing times, there will be small fluctuations in ...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Patents(China)
IPC IPC(8): G06K9/62
CPCG06F18/24
Inventor 吴炜占鸣王皓敖阗陈勇张殿斌孟然
Owner XIAN BANKNOTE PRINTING