Real-time, full web image processing method and system for web manufacturing supervision

Patent metadata

Field	Value
Publication number	US-10607333-B2
Application number	US-201816481896-A
Country	US
Kind code	B2
Filing date	Feb 9, 2018
Priority date	Feb 10, 2017
Publication date	Mar 31, 2020
Grant date	Mar 31, 2020

Abstract

A real-time, full web image processing method for analyzing formation in a web is described, the web is transported in a moving direction during a web manufacturing process, the method including the steps of acquiring a two-dimensional original image P0 of the web, the image being representable as a digital image representable by a plurality of pixel values P0,i,j with i∈{1; . . . ; I}, j∈{1; . . . ; J}; and producing a plurality of P processed images Pp with p∈{1; . . . ; P}, each of the processed images being representable by pixel values Pp,m,n with m∈{1; . . . ; M}, n∈{1; . . . ; N}, the processed images being obtained by spatial bandpass filtering of the original image, wherein a spatial different bandpass filter is used for obtaining each of the processed.

First claim

The invention claimed is:

1. A method, implemented on a computer, for detecting, monitoring and/or analyzing a quality of a product being produced in a manufacturing process, said product being transported, on a conveyor belt, in a moving direction during said manufacturing process, the method comprising the steps of:
a) acquiring an original image P 0 of the product, said image being representable as a two-dimensional digital image having a plurality of pixels having pixel values P 0,i,j with i∈{1; . . . ; I}, j∈{1; . . . ; J};
b) producing a plurality of P processed images P p with p∈{1; . . . ; P}, each of said processed images being representable by pixel values P p,m,n with m∈{1; . . . ; M}, n∈{1; . . . ; N}, said processed images being obtained by spatial filtering in form of spatial bandpass filtering, of the original image, wherein a different spatial filter in form of spatial bandpass filter, is used for obtaining each of the processed images, and
c) combining at least two, preferably all, of the processed images P p with p∈{1; . . . ; P} to obtain a feature map F being representable by values F m′,n′ with m′∈{1; . . . ; M′}, n′∈{1; . . . ; N′}, preferably m′∈{1; . . . ; M}, n′∈{1; . . . ; N} wherein the values F m′,n′ of the feature map correspond to a predominant size category for m′,n′ with F m′,n′ =F(m′,n′) p max,m′,n′ with P(p max,m′,n′ ,m′,n′)>P(p,m′,n′) with p∈{1; . . . ; P}\{p max,m′,n′ }, wherein the processed images P p with p∈{1; . . . ; P} are thresholded and converted to binary images representable by pixel values P p,m,n ∈{0; 1} for p∈{1; . . . ; P}, m∈{1; . . . ; M}, n∈{1; . . . ; N}.

2. The method according to claim 1 , wherein the product is a web, in particular a paper web, and the quality being monitored and/or analyzed includes formation in said web.

3. The method according to claim 1 , wherein the two-dimensional digital image by which the original image may be represented is provided as a stream of data, preferably in real time, and preferably without intermediate storage of the entire two-dimensional digital image.

4. The method according to claim 1 , wherein at least one of the plurality of P processed images P p is provided as a stream of data, preferably in real time, and preferably without intermediate storage of said processed image.

5. The method according to claim 1 , wherein characteristics of at least one spatial filter may be adapted, in particular by setting a filter parameter; and
a) the pixel values P p,m,n are obtained successively by applying one or more of the filters successively to individual pixels or subsets of pixels representing the original image; wherein
b) at least some pixel values P p,m,n that have already been obtained are used to adapt the characteristics of the at least one spatial filter prior to obtaining further pixel values.

6. The method according to claim 1 , wherein the feature map F m′,n′ is obtained according to
F m′,n′ P max,m′,n′ with P max,m′,n′ =max{P p,m′,n′ |p∈{1; . . . ;P}}.

7. The method according to claim 6 , wherein the feature map F m′,n′ is a scalar feature map, a first component of F m′,n′ contains values P max,m′,n′ , while a second component contains values P min,m′,n′ with P min,m′,n′ =min{P p,m′,n′ |p∈{1; . . . ; P}}.

8. The method according to claim 1 , further comprising the step of:
a) determining, from at least two, preferably all of the processed images P p with p∈{1; . . . ; P}, an image feature vector v=(v 1 , . . . , v P ), wherein vector component v p of said image feature vector v is determined from processed image P p with p∈{1; . . . ; P}.

9. The method according to claim 8 , further comprising determining the image feature vector v on the basis of the feature map F.

10. The method according to claim 8 , further comprising determining a first global image feature vector v on the basis of the whole original image and a second local image feature vector on the basis of a subregion or subarea of the original image and comparing the first and second image feature vectors.

11. The method according to claim 8 , further comprising the step of applying gain and/or offset correction to at least a selection of processed images P p with p∈{1; . . . ; P}, in particular applying individual gain and/or offset correction to a selection of processed images P p with p∈S⊂{1; . . . ; P}, wherein gain correction and/or offset for processed images P p is repeatedly updated based on a deviation between a current value of a local or image feature vector component v p and a target value {circumflex over (v)} p for said feature vector component v p .

12. The method according to claim 1 , wherein
a) the two-dimensional original image is obtained from a raw digital image of product web, preferably obtained by means of a real-time linescan or matrix camera using fixed scan time, and
b) said raw digital image is corrected by an adaptive flat line correction method.

13. The method according to claim 1 , wherein
a) in step b) of claim 1 , a plurality of smoothed images B q with q∈{1; . . . ; Q} each of said smoothed images being representable by pixel values B q,m,n , with m∈{1; . . . ; M}, n∈{1; . . . ; N}, is produced, each of said smoothed images being obtained applying a spatial low pass or smoothing filter to the original image, with a different filter being used for each of the smoothed images B q,m,n ;
b) each of the processed images P p with p∈{1; . . . ; P} is produced by subtracting two smoothed images B p1,m,n , B p2,m,n with p1≠p2.

14. The method according to claim 1 , wherein
a) a standard deviation σ of the original image P 0 is determined;
b) the processed images P p with p∈{1; . . . ; P} are thresholded with the standard deviation σ or a multiple thereof.

15. The method according to claim 14 , further characterized in that the feature map F is displayed as a two-dimensional digital color image, with a different color being displayed for each different value of F m′,n′ with m∈{1; . . . ; M′}, n∈{1; . . . ; N′}.

16. The method according to claim 1 , further comprising the step of displaying the feature map F as a two-dimensional digital image.

17. The method according to claim 1 , wherein at least one bandpass filter is a two-dimensional bandpass filter having transfer characteristics for a first spatial direction which are different from transfer characteristics for a second spatial direction.

18. An optical web inspection system comprising:
a) an image acquisition unit for acquiring a raw image and/or an original image P 0 of a web being transported in a moving direction during a web manufacturing process,
b) a digitization unit, preferably included by the image acquisition unit, for determining pixel values P 0,i,j with i∈{1; . . . ; I}, j∈{1; . . . ; J} representing said original image P 0 ,
c) a processing unit configured to execute the method including the steps:
a) acquiring an original image P 0 of the product, said image being representable as a two-dimensional digital image having a plurality of pixels having pixel values P 0,i,j with i∈{1; . . . ; I}, j∈{1; . . . ; J};

b) producing a plurality of P processed images P p with p∈{1; . . . ; P}, each of said processed images being representable by pixel values P p,m,n with m∈{1; . . . ; M}, n∈{1; . . . ; N}, said processed images being obtained by spatial filtering in form of spatial bandpass filtering, of the original image, wherein a different spatial filter in form of spatial bandpass filter, is used for obtaining each of the processed images, and
c) combin

Assignees

• Abb Schweiz Ag

Inventors

• Huotilainen Tommi

Classifications

• G06T2207/20024

  Filtering details · CPC title

• G01N21/89

  in moving material, e.g. running paper or textiles (G01N21/90, G01N21/91, G01N21/94 take precedence) · CPC title

• G06T7/0004 Primary

  Industrial image inspection · CPC title

• G06T5/20

  using local operators · CPC title

• G06T2207/30124

  Fabrics; Textile; Paper · CPC title