Title: Process for evaluating data from textile fabrics
Abstract: A process is disclosed for evaluating data obtained from textile fabrics. In order to devise a process which allows data obtained from textile fabrics to be easily compared, assessed in a differentiated manner as to their significance and evaluated, the data are determined in a section (3a,3b) of the surface of the fabric, sorted according to at least two parameters (13,14) and represented in an image (12, 30) as a function of the parameters.
Patent Number: 6,987,867 Issued on 01/17/2006 to Meier, et al.
| Inventors:
|
Meier; Rudolf (Uster, CH);
Uhlmann; Jürg (Frauenfeld, CH)
|
| Assignee:
|
Uster Technologies AG (Uster, CH)
|
| Appl. No.:
|
508430 |
| Filed:
|
August 14, 1998 |
| PCT Filed:
|
August 14, 1998
|
| PCT NO:
|
PCT/CH98/00343
|
| 371 Date:
|
March 13, 2000
|
| 102(e) Date:
|
March 13, 2000
|
| PCT PUB.NO.:
|
WO99/14580 |
| PCT PUB. Date:
|
March 25, 1999 |
Foreign Application Priority Data
| Current U.S. Class: |
382/111 |
| Current Intern'l Class: |
G06K 9/00 (20060101) |
| Field of Search: |
382/111
|
References Cited [Referenced By]
U.S. Patent Documents
| 3613743 | Oct., 1971 | Sakamoto.
| |
| 4728800 | Mar., 1988 | Surka.
| |
| 4745555 | May., 1988 | Connelly et al.
| |
| 5544256 | Aug., 1996 | Brecher et al.
| |
| 5745365 | Apr., 1998 | Parker.
| |
| 5834639 | Nov., 1998 | Meier et al.
| |
| 6100989 | Aug., 2000 | Leuenberger.
| |
| 6501086 | Dec., 2002 | Leuenberger.
| |
| 2002/0062775 | May., 2002 | Hoeller.
| |
| Foreign Patent Documents |
| 36 39 636 | May., 1988 | DE.
| |
| 0491954 | Jan., 1992 | EP.
| |
| 0 491 954 | Jul., 1992 | EP.
| |
| 2 701 766 | Aug., 1994 | FR.
| |
| 2 253 697 | Sep., 1992 | GB.
| |
Other References
B. Nickolay, et al., "Automatic Textile Inspection", Studie fur die Textilindustrie,
60 pages, Jun. 1993.
E. Ersö, et al., "Rationelle 100% Optische Kontrolle in Der Vlies—Und
Textilstoffproduktion", ISRA Systemtechnik GmbH, 16 pages.
Ferdinand van der Heijden, "Statistical Patter Classification and Parameter Estimation",
Image Based Measurement Systems, Object Recognition and Parameter Estimation, John
Wiley & Sons, 1994.
|
Primary Examiner: Johns; Andrew W.
Assistant Examiner: Nakhjavan; Shervin
Attorney, Agent or Firm: Buchanan Ingersoll PC
Claims
What is claimed is:
1. A method for representing faults detected on textile fabrics for purposes
of evaluation, comprising the following steps:
receiving data associated with a plurality of faults detected on a swatch of
the surface of the fabric;
sorting the data associated with the plurality of detected faults in accordance
with at least two parameters, wherein at least one of said two parameters pertains
to the size of the detected faults;
representing the received and sorted data associated with the plurality of detected
faults in an image having at least two dimensions, wherein one of said dimensions
corresponds to said one of said two parameters, and another of said dimensions
corresponds to the other of said two parameters.
2. The method of claim 1, wherein said one parameter is the length of a detected fault.
3. The method of claim 2 wherein the other of said two parameters is the intensity
of a detected fault.
4. The method of claim 2 wherein the other of said two parameters is the width
of a detected fault.
5. The method of claim 1 wherein said one parameter is the area of a detected fault.
6. The method of claim 5 wherein the other of said two parameters is the intensity
of a detected fault.
7. The method of claim 1 wherein said one parameter is the number of unit fields
in said swatch within which a detected fault is located.
8. The method of claim 7 wherein the other of said two parameters is the intensity
of a detected fault.
9. The method of claim 1 wherein said other dimension is divided into a plurality
of zones that are respectively associated with different types of faults.
10. The method of claim 1 wherein each of said two dimensions is divided into
a plurality of sections to thereby divide said image into a plurality of classes,
and the plurality of detected faults are represented as numerical values within
the classes with which they are respectively associated.
11. The method of claim 10 wherein the position of a numerical value within a
class indicates the value of a parameter for detected faults represented by that number.
12. The method of claim 11 wherein the parameter depicted by the positions of
the numerical values is a third parameter different from said two parameters.
13. The method of claim 12 wherein said third parameter is intensity.
14. A method for representing faults detected on textile fabrics for purposes
of evaluation, comprising the following steps:
receiving data associated with a plurality of faults detected on a swatch of
the surface of the fabric;
sorting the data associated with the plurality of detected faults in accordance
with at least two parameters, wherein at least one of said two parameters pertains
to the intensity of the detected faults;
representing the received and sorted data associated with the plurality of detected
faults in an image having at least two dimensions, wherein one of said dimensions
corresponds to said one of said two parameters, and another of said dimensions
corresponds to the other of said two parameters.
15. The method of claim 14 wherein one of said dimensions is divided into a plurality
of zones that are respectively associated with different types of faults.
16. The method of claim 14 wherein each of said two dimensions is divided into
a plurality of sections to thereby divide said image into a plurality of classes,
and the plurality of detected faults are represented as numerical values within
the classes with which they are respectively associated.
17. The method of claim 16 wherein the position of a numerical value within a
class indicates the value of a parameter for detected faults represented by that number.
Description
FIELD OF THE INVENTION
The invention relates to a method for evaluating data determined on textile fabrics.
BACKGROUND OF THE INVENTION
When producing textile fabrics such as woven fabrics, knitted fabrics, etc.,
faults which cause the ideally regular and precisely structured surface to exhibit
irregularities or faults are a frequent occurrence. In terms of extent, faults
of this kind may range from being very small and inconspicuous to very large or,
for other reasons, conspicuous and may reduce the value and the function, e.g.
the strength or the appearance of the fabric. The finished fabrics are therefore
subjected to an examination for the purpose of indicating faults in the structure.
This may be a visual or a machine examination and often takes place both before
dyeing or dressing and also before making up. An increase in the quantity of detected
faults is to be expected in particular when carrying out a machine or automated
examination, so that a correspondingly greater data flow may result.
One disadvantage in this case lies in the fact that, although a considerable
amount of data is available, these data are likely to cause confusion and may not
just serve to improve the quality of the products. It should also be borne in mind
that there are a great many producers of textile fabrics of all kinds and that
each producer and also many customers are inclined to define and implement their
own quality criteria. This means that textile fabrics which are assessed by different
individuals or institutions result in assessments which cannot easily be compared
with one another.
SUMMARY OF THE INVENTION
As characterized in the claims, the invention therefore achieves the object of
providing a method by which faults which are determined in textile fabrics can
easily be compared with one another and assessed and evaluated as to their significance
in a differentiated manner.
This is achieved by determining the data on a swatch of the surface of the fabric
and sorting this data according to at least two parameters. A swatch can be understood
to be the entire surface under consideration of a fabric or a section from the
surface. A section of this kind may be moved or changed after a period required
for acquiring the data, so that new data on other zones or swatches of the fabric
are periodically obtained. The intensity of a pixel or surface element, a longitudinal
coordinate, a latitudinal coordinate, etc. may be considered as data and therefore
also as parameters, for example. The acquired data on the faults are then represented
in an image as a function of selected parameters, which in turn may be divided
into zones which in themselves are conceived as homogeneous. If two parameters
are selected, the result is a one-dimensional representation. If three parameters
are selected, the resulting image is a two-dimensional representation. The image
then represents, for example, a classifying field consisting of individual fields
which define a class. The class is characterized by the extent of the field, which
lies in a plane which is regarded as the location for values of two parameters.
A further parameter may be displayed by symbols entered in the field.
The advantages achieved by means of the invention lie in particular in the fact
that it enables a structured and standardized assessment of faults in textile fabrics
to be carried out. Thus on the one hand values of predetermined parameters for
the most varied faults can be indicated, while on the other criteria can be created
which help to identify the significance or value of the faults and to compare this
with the value of other faults. A large data flow on faults in the fabrics can
thus also be processed to provide accurate information on the faults occurring.
BRIEF DESCRIPTION OF THE DRAWINGS
The invention is illustrated in detail in the following on the basis of an example
and with reference to the accompanying figures, in which:
FIG. 1 shows a respective swatch of the surface of a textile fabric,
FIG. 2 shows a respective swatch according to FIG. 1 with different faults, and
FIGS. 3 to 11 in each case show a classifying field.
DETAILED DESCRIPTION
FIG. 1 shows the same run
1 of a textile fabric three times with a fault
2. Information on the position of this fault
2 can be obtained, for
example, via coordinates x and y, on its size via values of the extent in two directions
s and k, and on its intensity or deviation, for example in terms of color, from
the surrounding area via a value delta i.
FIG. 2 shows a respective swatch
3a,
3b of a textile
fabric with a grid
4 and four different faults
5,
6,
7
and
8. The swatch
3a shows a first possibility for evaluating
the size of the faults
5,
6,
7 and
8 and the swatch
3b a second possibility. For this purpose the grid
4 divides
the swatches
3a,
3b into individual small fields
9,
and the occupancy of these fields by the faults
5-
8 is interpreted
differently in the two swatches
3a and
3b, as will
be discussed further in the following. However in both cases this means that the
extent of the faults through the number of occupied fields is selected as a parameter.
Although—should this be a woven fabric—the faults
5,
6,
7,
8 extend in two directions, weftwise
10 and warpwise
11,
the values of the parameters only indicate that the intensity of the faults
5-
8
has exceeded a threshold value and one of the number of occupied fields
9
has a proportional extent. The swatches
3a,
3b preferably
form at least one rectangle whose sides extend parallel and perpendicularly to
boundaries of the fabric or run
1.
FIG. 3 shows an image
12 with two axes
13,
14, along which
values of parameters are plotted. Here the values along the axis
13 are
values for the length of a fault, for example viewed weftwise in a woven fabric,
and those along the axis
14 values for the width of a fault, for example
viewed warpwise in a woven fabric. Lines
15,
17,
19 and
21
divide the width of the faults into five classes, while lines
16,
18,
20 and
22 divide the length of the faults into five classes. This
results overall in twenty five classes for classifying the faults according to
size. Symbols
23-
29 are drawn in at a plurality of class boundaries,
which are indicated by the lines
15-
22, these symbols representing
the form of a fault as is to be expected on the basis of dimensions according to
the said lines. Numerical values are also entered in the fields defined by the
lines
15 to
22, these values indicating the number of detected faults
which fall within the class concerned. For this purpose it is assumed that a class
represents a homogeneous zone, i.e. no distinction is made as to whether or not
the values of the parameters lie near upper or lower class boundaries or lines
15-
22.
FIG. 4 shows an image
30 with axes and lines defining classes as is already
known from FIG. 3. The axes, lines and symbols have therefore been given the same
reference numbers. Dots
31,
32,
33, etc. are entered in the
fields, the position of which dots in relation to the axes
13 and
14
indicates the size of the fault accurately or in a differentiated manner. Each
dot therefore corresponds to a fault, and the distribution of the faults or the
dots thereof is also an indication of the predominant type of fault in the fabric.
Characters A to E are also entered along the axis
13 between the lines
14
to
22 and integral numbers 1 to 5 along the axis
14 between the lines
13 to
21. Each field and therefore each class can therefore be clearly
designated by the combination of a number and a character. FIG. 5 shows an image
34 with axes and lines defining classes as is already known from FIG. 3.
The axes, lines and symbols have therefore been given the same reference numbers.
Diagonally ascending numerical values, which indicate the intensity of a fault,
are provided in the individual fields, which correspond to fault classes. Here
the position of a figure indicates the intensity, while the value of the figure
indicates the number of faults with this intensity. Thus numerical values located
in the bottom left-hand side of a field indicate high intensities and numerical
values located in the top right-hand side indicate low intensities. FIG. 6 shows
an image
35 with axes
36 and
37. Values for the area of a
fault, for example in CM
2, are plotted along the axis
36 and values
for the intensity of a fault in percentages along the axis
37. This image
35 is also divided into fields or classes by lines
38 to
43.
Symbols which indicate the intensity of the fault through the strength of the color
are drawn in at the intersections of the lines
38-
43. Numerical values
in the fields indicate the number of faults occurring in the class concerned.
FIG. 7 shows an image
44 with axes
45 and
46. Values for
the length of a fault, for example in cm, are plotted along the axis
45
and values for the intensity of a fault, for example in percentages, along the
axis
46. This image
44 is also divided into fields or classes by
lines
47 to
52. The number of detected faults is indicated by the
figures in the fields, as already known from FIG. 3. FIG. 8 shows an image
53
with axes
54 and
55. Values for the number of occupied fields
9
according to FIG. 2 are plotted along the axis
54 and values for the intensity
of a fault along the axis
55. This image
53 is also divided into
fields or classes by lines
56 to
61. The number of detected faults
is indicated by the figures in the fields, as already known for FIG. 3.
FIG. 9 shows an image
62 with axes
63 and
64. Values for
the length of faults in cm are plotted along the axis
63. The axis
64
is divided into a plurality of zones
64a to
e, and values
for the intensity are given in percentages in each zone. Each of the zones
64a
to
64e relates to a certain type of fault, for example the zone
64a relates to weft faults, the zone
64b to warp faults,
the zone
64c to surface faults, the zone
64d to edge
faults and the zone
64e to holes. Lines
65 to
76 again
divide the image
62 into fields or classes in which numerical values indicate
the number of detected faults in the class concerned. The position of the numerical
value in relation to the zone on the axis
64 indicates the intensity of
the fault. Several numerical values may thus also occur in one class. The image
62 thereby illustrates a classification which is based on different types
of fault. Different known types of fault may be grouped together as desired. So,
for example, the term "weft faults" is here generally understood to mean faults
which predominantly extend weftwise in a woven fabric. Such faults are known under
the following terms: join, fell, straightening point, shed, weft bar, lashing-in,
slubber, fly, thread breakage, mispick.
FIG. 10 shows an image
80 with an axis
81 which is divided into
zones
81a to
d. Values for intensities in percentages are
given along another axis
82. Lines
83 to
93 divide the image
80 into fields or classes. Values for the number of detected faults can
again be entered in the fields or classes. For example, the intensity of weft faults
can be entered in zone
81a, the intensity and size of wrap faults
in zone
81b, the intensity or size of holes in zone
81c,
the intensity of edge faults, etc. in zone
81d, and the numbers thereof.
FIG. 11 shows an image
94 with axes and lines as already found in images
12 and
30 (FIGS. 3 and 4). Here the fields or classes are divided
by a boundary
97 into two groups
95 and
96, with the boundary
extending along lines
15,
17,
19 and
16,
18,
20. However it is also possible to define a boundary
98 which also
divides the individual fields or classes.
The method according to the invention is carried out as follows: The textile
fabric is scanned in a manner known per se, for example by a camera, and images
for swatches of the surface of the fabric are made and signals derived therefrom
are processed. Using algorithms, which do not constitute the subject matter of
this invention, for image processing, faults or unusual features in the images
of the surface are determined from the derived signals by comparison with predetermined
limit values, patterns, etc. Thus data on faults in a swatch of the fabric are
produced. A swatch of this kind is shown, for example, in FIG. 1 and called a run
1. A fault
2, which is distinguished by various parameters, can be
recognized in this. These parameters are its position, which is given by coordinates
x and y, its size, which is given by the values s and k, and its intensity, which
causes the fault to actually stand out from the area surrounding it and which is
quantified by a qualitative datum, here called delta i.
Different parameters are significant, according to how the fault is subsequently
dealt with. For example, if every fault is to be removed, all that is of interest
is its position, possibly also its size. If the fabric is then to be assessed as
to where the faults are most numerous, such as at the edge, for example, it is
again just the position which is of interest. The data are then sorted according
to parameters such as length and width and accordingly represented in an image.
Should there be a requirement for assessing how the fault appears to the eye
or how it influences subsequent processing of the fabric, such as dyeing or dressing,
its size is of interest and possibly also its intensity. Then the parameters according
to which the data are sorted are the length s and the width k of the fault, as
well as its intensity delta i.
Just one dimension may be determined from the signals obtained from image processing
in order to detect the size of a fault, or an evaluation according to FIG. 2 may
be undertaken. In this case an investigation is carried out to establish how many
fields
9 are affected or at least partly covered by a fault. These fields,
as marked in swatch
3a, are counted for each fault and the number
is plotted, for example, along the axis
54 in FIG. 8. However it is also
possible, as shown for swatch
3b, to take the fields
9 occupied
for each fault and to complete them to an extent such that together they form a
rectangle which encompasses the fault. The fields
9 which are comprised
in this rectangle then have to be counted and plotted.
In order to detect the intensity of a fault, the color or brightness of the area
surrounding the fault is taken as a starting point and an attempt is made to quantify
deviations of the color or brightness more or less accurately or in a graduated
manner, this being expressed by a value delta i. The devices used for image processing
determine the degree to which this is successful.
In order to represent the size of the fault in an image, its length can be detected
in the swatch in a manner known per se and represented in an image
12,
30
by a value on the axis
13. The width of the fault can be represented in
the same way by a value on the axis
14. Together these two values produce,
for example, a dot
33 (FIG. 4). This can be left as a dot or simply treated
as a fault in class C
2, which would mean that just one counting value would
then be increased by one for this class. For this purpose it is possible to specify
certain fields or classes as acceptable and others as unacceptable beforehand.
The position of the fault in image
13,
30 then immediately reveals
how the fault is to be assessed. Should values for faults accumulate in individual
classes, this will equally provide an indication for assessing the fabric.
The intensity of a fault can be represented according to the possibilities already
presented on the basis of the images
34,
35,
44 and
53
(FIGS. 5-8).
As shown in FIG. 1, swatches of the surface from which the data are acquired
which
form a rectangle are particularly suitable, for the fabrics in question are also
already in the form of rectangles, this being a result of the manufacturing process.
Then sides of the swatches should also lie parallel and perpendicularly to the
boundaries of the fabric. However the swatch concerned does not conventionally
constitute the entire surface of the fabric. This applies to swatches
3a,
3b according to FIG. 2, which is an enlarged view of a part of the
run
1 according to FIG. 1.
The form of a fault, as represented by the symbols
23 to
29 in
FIG. 3, may also be directly considered as a parameter. In fact a parameter of
this kind ultimately consists of two parameters (length and width). However it
would also be possible to combine the parameter "form" with the parameter "intensity",
as known from FIG. 6, and in this way obtain another combination and therefore
another image representation. It thus becomes obvious that only a few possibilities
are indicated here, although these can also be developed according to the invention
in an obvious manner by combination, for example by interchanging the axes.
Data can be evaluated and, optionally, the textile fabric processed in a differentiated
manner, according to whether the determined data belong to groups
95 or
96 (FIG. 11), which are separated by a boundary
97,
98. For
example, the weighting of the faults in group
96 may be reduced with respect
to the faults in group
95. Or faults of group
96 are only marked,
for example, at the edge of a cloth run, while faults of group
95 are removed,
for example by unraveling the woven fabric in the area around these faults. Generally
speaking, boundaries
97,
98, etc. can form groups of classes or categories
of faults which initiate different actions.
*
