WO2004057317A1 - A method for classifying the quality of a surface - Google Patents

A method for classifying the quality of a surface Download PDF

Info

Publication number
WO2004057317A1
WO2004057317A1 PCT/DK2003/000923 DK0300923W WO2004057317A1 WO 2004057317 A1 WO2004057317 A1 WO 2004057317A1 DK 0300923 W DK0300923 W DK 0300923W WO 2004057317 A1 WO2004057317 A1 WO 2004057317A1
Authority
WO
Grant status
Application
Patent type
Prior art keywords
values
surface
constitution
computer
reference
Prior art date
Application number
PCT/DK2003/000923
Other languages
French (fr)
Inventor
Ulrik SCHØNNEMANN
Visti Andresen
Martin Carlsen
Original Assignee
Aminova Vision Technology Aps
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date

Links

Classifications

    • GPHYSICS
    • G01MEASURING; TESTING
    • G01NINVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00Investigating or analysing materials by the use of optical means, i.e. using infra-red, visible or ultra-violet light
    • G01N21/84Systems specially adapted for particular applications
    • G01N21/88Investigating the presence of flaws or contamination
    • G01N21/89Investigating the presence of flaws or contamination in moving material, e.g. running paper or textiles
    • G01N21/892Investigating the presence of flaws or contamination in moving material, e.g. running paper or textiles characterised by the flaw, defect or object feature examined
    • G01N21/898Irregularities in textured or patterned surfaces, e.g. textiles, wood
    • G01N21/8986Wood
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01NINVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00Investigating or analysing materials by the use of optical means, i.e. using infra-red, visible or ultra-violet light
    • G01N21/84Systems specially adapted for particular applications
    • G01N21/88Investigating the presence of flaws or contamination
    • G01N21/8851Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
    • G01N2021/8854Grading and classifying of flaws
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01NINVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00Investigating or analysing materials by the use of optical means, i.e. using infra-red, visible or ultra-violet light
    • G01N21/84Systems specially adapted for particular applications
    • G01N21/88Investigating the presence of flaws or contamination
    • G01N21/8851Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
    • G01N2021/8887Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges based on image processing techniques
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01NINVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00Investigating or analysing materials by the use of optical means, i.e. using infra-red, visible or ultra-violet light
    • G01N21/84Systems specially adapted for particular applications
    • G01N21/88Investigating the presence of flaws or contamination
    • G01N21/8851Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01NINVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N2201/00Features of devices classified in G01N21/00
    • G01N2201/12Circuits of general importance; Signal processing
    • G01N2201/127Calibration; base line adjustment; drift compensation
    • G01N2201/12746Calibration values determination

Abstract

A method for classifying the quality of a surface (5) of a material (4) in relation to at least one reference constitution (13) comprises the steps of; storing in a computer reference values (27) representing the at least one reference constitution; detecting the constitution of the surface; reading in the results of the detection as input to the computer; converting the input to sets of part values (23) representing each a part of the surface; comparing said values with the reference values (27) stored in the computer; reading out the comparisons as output from the computer; and classifying the surface (5) quality of the material in accordance with said output. Thereby a method advantageously is obtained in which surface (5) deviations quickly and easily are detected on many different types of surfaces by using a method that is easily adapted to be used with different detection devices (8) and in different applications and industries.

Description

A method for classifying the quality of a surface

The present invention relates to a method for classifying the quality of a surface of a material in relation to at least one reference constitution.

Apparatuses and methods for scanning surfaces on materials and thereby determining the quality of the material are widely used in productions where the quality of the final product is important.

Many such apparatuses and methods are described in the art, such as the American patent US 6,122,065 where an apparatus and a method for detecting surface shapes such as roughness, cavities, wane, missing wood and altered wood on lumber pieces which are conveyed at high speed. However this invention are mainly suited for detecting the mentioned characteristic on wood and the devices for detecting is limited to lasers and cameras .

The European patent application 1 092 973 tries to overcome the above, but is still limited to an imaging device as the detecting device.

Further the processing of the received data is also very specific in the known art, which can be seen in US 5,274,714 where it is limited to Fourier transform.

There exist today a growing demand for a surface detecting method, which can be implemented with different types of detection devices for detecting a variation of material surfaces. Especially in larger production companies where many different types of materials and quality tests are necessary a general method for providing surface inspecting are needed in order to save the costs of buying many different systems for each type of material and detection type and also having to train workers in the use of these many different systems.

It is therefore an object of the present invention to provide a new and improved method to inspect surfaces of materials so that it can be implemented in various productions and with various detecting devices .

According to the invention a method for classifying the quality of a surface of a material in relation to at least one reference constitution comprises the steps of; storing in a computer reference values representing the at least one reference constitution; detecting the constitution of the surface; reading in the results of the detection as input to the computer; converting the input to sets of part values representing each a part of the surface; comparing said values with the reference values stored in the computer; reading out the comparisons as output from the computer; and classifying the surface quality of the material in accordance with said output .

Thereby a method advantageously is obtained in which surface deviations quickly and easily are detected on many different types of surfaces by using a method that is easily adapted to be used with different detection devices and in different applications and industries .

In a preferred embodiment of the invention the computer has a program for advantageously classifying the material in accordance with detected surface deviations by calculating values for the deviation of the part values from reference values, representing at least one first reference surface constitution; values for the grading of the part values into classes in accordance with predetermined criteria for the deviations; values for a number of total values, which are representing each a group of a predetermined minimum number of part values for the constitution of abutting and/or closely spaced surface parts having part values in the same class; values for the deviation of said total values from reference values representing at least one second reference surface constitution, ; values for the grading of the total values into classes in accordance with other predetermined criteria for the deviations; at least some of the calculated values can be read out as output from the computer, and the surface quality of the material can be classified in accordance with said output.

Thereby an advantageous embodiment of the method is obtained where the deviations on a surface easy and accurately can be classified into groups, which represent one type of configuration.

Further in a method according to the invention the results of the detecting of the constitution of the surface can be read into the computer in form of digital signals, representing each the minimum surface area, which can be detected with a detecting device having a given resolution; the digital signals can be converted to values representing each one of said minimum areas; assembling a number of said values to sets of part values representing each a part of the surface; at least some of the calculated values can be converted to digital signals; and said digital signals can be read out as output from the computer.

Thereby a signal will be generated containing only the most relevant information, which will improve the speed of any successive processing.

By an advantageously embodiment according to invention the surface parts can be classified by predetermined criteria for deviations, and the groups can be classified by predetermined criteria for number of groups per unit of surface to be detected. Thereby the classification easily can be changed and applied to the relevant material and the demanded quality of that material .

To improve the overview and the speed of the method the groups can advantageously be arranged in a matrix, which represents the surface mapping into the groups, and the matrix can be read out as signals representing the positioning of said groups on the surface.

In a preferred embodiment the constitution of the surface can be determined by a statistical classification of the total values, whereby the total values can be classified in a fast, reliable and safe way.

A more specific example on such a statistical classification will be provided in the detailed description.

Advantageously the computer can be taught the reference constitution, this can be done in a simple and efficient way by steps for selecting a material with a surface constitution, which corresponds to the at least one reference constitution; detecting the surface constitution of the material; reading into the computer the detected result in form of digital signals; and converting the digital signals to reference values; and storing the reference values.

Thereby an easy and quick method for providing the computer with the reference values are provided in a way that is unambiguous to any kind of production line.

In a preferred embodiment the first values chosen to represent the constitution can be chosen from a large group of known first values. Thereby the efficiency, both in terms of speed and reliability are greatly improved. A specific algorithm choosing the first values from a known group of first values will be provided in the detailed description.

To visually illustrate the detected surface and the deviations to a user of an apparatus with the method implemented the output read out from the computer can be a graphical representation on a display or a printout on a sheet of paper with indication of at least the positioning of the groups and the classes of the constitution of surface parts of which the groups are consisting. Alternatively the output can be in shape of signals representing the classes of the groups and the surface parts of these.

Many different kinds of detecting devices can be used for detecting the surface and the detection of the surface. The scanning can for example be carried out by means of laser scanning, ultrasound scanning or photography by for example an x-ray camera, a CCD-camera or a CMOS-camera.

In a preferred embodiment the material can be sorted in dependence of the quality of the surface by using the method according to the invention in such a way that the material can is sorted in accordance with the classes read out as output from the computer.

The method can for example be used for sorting and/or scrapping materials such as wood, metal, plastic, cloth, paper, and/or such materials that are painted and/or printed with a picture and/or an inscription.

It shall be understood that the term surface as used herein not should be restrictive in the sense that it is only the visible two-dimensional planar top part of a material. The term surface also includes a region below this planar top part, which can be detected by a known device. The term likewise covers the a structure with deviating thickness in respect to a reference point .

The invention will be explained in greater details below, describing only simple exemplary embodiments with reference to the drawing, in which,

Figure 1 shows an apparatus for carrying out the method according to the invention in a perspective view,

Figure 2 shows a section of the apparatus taken along the lines II-II in fig. 1 in a schematic view,

Figure 3 shows a flowchart of a method according to the invention,

Figure 4 shows in a large scale schematically an incorrect dimensioned section area,

Figure 5 shows in a large scale schematically another incorrect dimensioned section area,

Figure 6 illustrates graphically the groups of features according to a formula of the invention, and

Figure 7 shows a flowchart of a learning process according to the invention.

Fig. 1 and 2 illustrate a typical apparatus for carrying out the method according to the invention

The apparatus comprises a housing 2 with a channel 3 for passage of a material 4 having four surfaces 5 to be classified. The material is by means of conveyor belts 6, arranged on each side of the housing, transported through the channel in the feed direction indicated by the arrow. The housing has four walls 7 parallel to the feed direction. On the inside of each wall a set consisting of a digital camera 8 and two light sources 9 is placed. The sets are separated from each other by shielding walls 10, which each is extending from an inside corner 11 of the housing to the diagonal opposite corner 12 of the housing.

During operation the material 4 will be fed through the channel of the housing in the feed direction shown while the constitutions 13 of the four surfaces 5 of the material are detected by each camera 8 placed on the wall 7 opposite the respective surface 5.

The light sources are arranged on each side of and slightly behind the lens of the camera to provide an even and constant ambient light, and the shielding walls 10 are arranged to prevent light sources on opposite walls from interfering with other cameras by emitting a direct light.

The detected surface captured by the camera will be converted to an electrical signal and transmitted to a computer for processing.

Such a computer could be arranged in the housing (not shown) and a control panel 14 on the outside of the housing could be arranged to provide a user with information or allow a user to read in input to the computer.

A typical digital camera is able to provide information on the colour of the surface 5 of the material and the intensity of the light reflected when known light sources are used as above .

As will be understood by a person skilled in the art such surface inspection apparatus as described above may be manufactured in many different ways. Further it is not restrictive for the method according to the invention that it is photography by a digital camera, such as a CCD or a CMOS- camera, that are used for detecting the surface, but many other methods can be used such as laser scanning, ultrasound scanning or photography by for example an x-ray camera.

Ultimately it is the type of constitution, and the method used for detecting that constitution that dictates the type of detecting device used, and it should be understood that the choice of the detecting device is not restrictive for the method according to the invention.

As example can be named a specific production where white plastic profiles 4 are extruded and these are inspected for dark spots 13 on the surface 5, which appear as a result of production errors.

As shown in fig. 1 and 2 only the four visible surfaces 5 of the plastic profiles 4 are inspected. Each surface has in this case a width of 150 mm. The profiles are inspected directly after extrusion (not shown) , which is performed at a maximum speed of in this case four meters per minute.

The spots to be detected have different intensity and diameter. It is estimated that the smallest spots relevant to the quality of the plastic profile are 0,1 mm in diameter.

For each surface there is used a camera setting with a minimum of 1.500 pixels to detect the smallest spots over the whole width. A standard grey tone camera is used since the spots only differ from the rest of the surface with respect to the intensity. A standard line-scanning camera known in the art with a frequency of 667 Hz and a resolution of 2048 pixels is advantageously used. The two light sources per camera are supplied by a DC source to avoid blinking interference to the camera, which would occur with an AC source.

With the above specification there is generated approximately 1MB of data every second from each surface.

For computation of the four 1MB channels a standard Pentium l,2Ghz personal computer with 512MB Ram is used.

Fig. 3 shows a flowchart describing the processing of the detected surface.

In step 15 one of the cameras detects the constitution 13 of the surface 5. The surface as detected by the camera 16 is divided into thirty-two pixels 17, which is the resolution of the camera. Said resolution is only a simplified example, as the resolution of a standard digital camera is much higher.

In step 18 the method converts the constitution 13 from the detection to values 20 representing the minimum surface area for the constitution 13. This is basically a standard analogue- to-digital conversion, where the analogue signal is converted to a digital signal. In case of using a typical digital camera, this will occur simultaneously with the detection of the surface by the digital camera.

However it is within the scope of the invention that the invention also could be used for detecting the surface of e.g. a paper photograph or other paper print, which then would have to be scanned to be able to be converted to a digital representation 19.

The digital representation 19 will in most computers known today basically be in a binary representation, where each pixel of the digital camera contains a value with a size such as 32bit. The resolution of the surface 5 is still determined by the resolution of the camera or as above mentioned other situation the resolution of the scanner.

The values representing the minimum surface area 20 will typically represent a measurable parameter depending on the device used for detecting the surface 5. As mentioned above the representative values 20 in the apparatus described in fig. 1 and 2 are representative for the colour in each pixel 17 and the intensity of the light. Other possible representative values could be diffraction of light, the time delay for a signal to be emitted and received, such as a laser beam, which emits pulses that is reflected by the surface of the material and returned to a detector, or the hardness of the surface measured by standard Vickers, Brinell or Rockwell hardness tests known in the art.

A representative value 20 can also be processed by a method known in the art in order to yield the proper information. Such a method could be a Fourier transform, typically a Fast Fourier Transform or other digital filters in order to yield specific mathematical values such as the roots of a curve.

In the embodiment described above and illustrated in fig.l and 2 the signal sent for processing would be the signal as represented right before step 21.

The resolution of the detected surface 16 is changed in step 21 and the representative values 20 are assembled into sets of part values 23. Each set of part values represents a part 24 of the detected surface 16.

The change in resolution is made in such a way that unnecessary computations can be avoided and the speed of the process is improved. The part of the surface 24 determines the new resolution and the dimensioning of these is described later in the document .

In step 25 the process compares the representative sets of part values 23 with reference values 27. For accomplishing that these reference values are read into the computer 26 they are determined in an earlier learning process, which will be described later, and are typical stored on a standard storage medium such as a hard drive, CD or DVD rom or even stored externally and retrieved over the Internet.

Based on step 25 the method will afterwards classify the representative sets of part values 23 into values for the deviation 31 in step 29.

This classification can be done in different ways known in the art. One such method is by statistical classification of the representative values 20.

In an advantageous embodiment the statistical classification can be performed by the formula :

PC(Z) e<*-*>τ∑-l<*- >'2 χ,μ e R" ∑ e R"xn n,c e N (32)

Figure imgf000013_0001
Where χ is the representative set of part values, Σ is the covariant matrix and μ is the mean vector for the group c.

The surface part 24 will now represent one of the constitutions that the material is categorized in. According to a predetermined set of criteria the process groups the values of the deviation 31 into groups 35 in step 33.

As illustrated one such set of criteria is, that all values of the deviation with the reference A 36 represent the reference constitution and are accepted by the process without further computations. Another criteria is that, two or more neighbouring values of the deviation with the reference B 37 are grouped into one group, thus the single standing reference B values of the deviation is ignored, yet a third criteria states that just one or more reference C 38 values of the deviation shall constitute a group 35.

Other examples on criteria for grouping the values of the deviation 31 could be that they were placed with a predetermined maximum number of surface parts between them or that they should only be grouped if they were placed a number of surface parts away from the periphery of the surface .

In the specific embodiment described above with the inspection of plastic profiles criteria are set up to detect spots with a specific diameter. A surface part 24 is thus defined by threshold values for the intensity of the light and if it is within a certain value the surface part is categorized as a group 35 describing spot. A group thereby represents a physical constitution 13 that is provided on the surface 5 of the material 4.

Since the minimum resolution requirements were 1.500 pixels and the line-scan camera used has 2.048 pixels, one single pixel 17 within the values for the intensity of spots is enough for the plastic profile to be discarded. This will however discard profiles with a minimum spot diameter of 0,07mm, if a spot diameter of 1,4mm could be acceptable a group of four pixels or more would define a spot. However, since this is the only criterion, the process could be optimised by defining a surface part 24 as consisting of four pixels, as described in step 21, whereby the processing time would be considerably reduced.

In another embodiment there are mapped criteria for sorting the material depending on the different groups 35 and their placement on the detected surface into predetermined classes . Such classes may be classes for scrap able materials, materials for use in other productions, recyclable material or second- quality material .

Final a signal representing the placement and type of groups is generated as an output in step 39. This output could be as a display signal on a computer monitor 40, which graphically shows the placement and types of groups 35.

Other outputs 36 such as to a printer or a storage medium could also be possible, and none of these or other output destinations is restrictive to the scope of the invention.

Further the output could go directly to a sorting machine (not shown) , which with use of for example robotics sorts the material into their respective classes.

The above method described with reference to fig. 3 can advantageously be implemented as a program where the representative values 20 are arranged in a matrix (not shown) which represent the surface 5 mapped into surface parts 24 and that the matrix is read out as signals which at least represents each of the groups 35 placement on the surface 5.

Thus the process is easily monitored and corrections to threshold values and the like can easily be adjusted.

To ensure that process works according to the invention the surface parts 24 must be defined such that they will contain the relevant values representing the minimum surface area 20.

In fig.4 a surface part that is defined to large 41 is shown, thereby containing two values of the deviation with reference A and B 36,37, which depending on the reference values 27 may be misinterpreted by the process as a value of the deviation with reference C 38. Ultimately this may lead to misinformation of the defect or deviation of the surface of the material and may in some instances lead to that actual acceptable material gets discarded.

Opposite to the situation described above the surface part may also be defined to small 42, which is illustrated in fig. 5.

Thus an area which is actually a value of the deviation with reference B 37 may be described as a partly an area with values of the deviation with reference A 36. However, this case will as such not be basis for any misinterpretation, but defining the surface parts smaller than necessary will however influence the processing speed, since more computations are required for a higher resolution of surface parts 24 and thereby the process will be slowed.

To avoid unnecessary computations and thereby also improving the speed of the method according to the invention only the representative values needed 20 to detect the values of the deviation 31 will be used.

As an example a user of the apparatus 1 described in fig. 1 and 2 may not actually need to both process the colour and the intensity of the light of a surface part to determine the different surface parts. Thus the unwanted representative value 20 will be discarded to improve the performance of the apparatus 1.

The above-mentioned issue is graphically illustrated in fig. 6 where a set 0 43 is the total amount of representative values

20 available to an apparatus according to the invention. These known representative values 20 can for example be part of a program, which is supplied along with devices 8 used for detecting the surface 5 or downloaded from the Internet. R 44 is a subset of 0 and contains the representative values 20 necessary to describe the different constitutions 13 of the surface 5 of the materials 4.

As can be understood this can save a considerable amount of processing time by the computer. Should a user, however, want to use a representative value 20 not contained in the subset R 44, the program used can advantageously allow the user to add this value to the process.

The subset R 44 may be determined in different ways. In one way it could be empirically determined or an algorithm could be implemented in a teaching phase of the apparatus 1 where R is determined.

In a preferred embodiment the following algorithm 45 is used:

Let D be the number of original representative values. Let 0 be the amount of original representative values; 0 = {Fi|i=l,...,D}. Let S(Fi,Fj) denote the degree of difference between representative values Fi and Fj , where the degree of difference is the maximum information compression index λ2: λ2{i ) = var(t) + var(y) - sqrt((vav(i)+ var( ))2 -4 var(t)var( )(l - p(i, j)2)) . Let rι denote the degree of difference between representative values Fi and its k closest neighbour in R, where R is a subset of 0, which is derived by the steps: a) Chose a starting value for, 0<k<=D-l. Let the starting value for R is 0. b) Calculate rkι for each representative values Fj. belonging to R. c) Find a representative values F'i, where rkι' is minimal. Keep F'i in R and remove the k closest representative values F'i from R. Let ε be ri' . d) If k>|R|-l:k=|R|-l. e) If k=l:go to h. f) If rki'> ε:

1 )

Figure imgf000018_0001
belongs R£" i ' ii) If k=l:go to h. iii) Go to 6. g) Go to b. h) Pick R to be the reduced number of representative values .

The above described method for determining a reduced number of representative values 20 are in a preferred embodiment of the invention part of a learning process, wherein the reference values 27 for the method is determined and also the size of the surface part 24 is determined, which was discussed above with reference to fig. 4 and 5.

In fig.7 the steps of the learning process are shown in a flowchart.

One or more materials are selected on which all types of constitutions are present.

In step 46 the surface 5 of the selected material 4 is detected and in step 47 the data is transmitted into a computer .

In step 46 areas with different surface structures are then marked. This is typically done in a painting program on a computer where the different areas are circled or similarly marked with different colours. The areas can then later be unambiguously identified from the digital representation of the surface whereon the marked areas are present .

In an alternative embodiment the areas can be marked before the materials are detected in step 46. This can be done by a standard colour marker, which will adhere to the surface of the material. Different constitutions are marked with different colours so that the computer later can separate the different constitutions.

In step 49 the data received in the computer is processed and the size of the surface parts 24, the representative values 20 needed to describe a constitution and the reference values 27 for the representative values 20 are determined.

By analysing the different marked areas on the material reference values 27 for the different representative values, representing the reference constitution for one type of material can advantageously be found and used for later surface detection.

The number of reference values 27 will be the same as the reduced number of representative values 20 as determined by the process described with reference to fig. 6.

The reference values 27 are advantageously sorted in a matrix

(not shown) where each reference value will refer to one or more constitution 13 depending on the size of the value. Thus a set of reference values 27 corresponding to a representative set of values 23 will be able to categorize the representative set of values 23 representing a surface part 24 to represent a specific constitution 13 on the surface 5 of the material 4. In step 50 these reference values 27 are stored on a storage medium, such as a hard drive.

After the learning process is complete the apparatus 1 is configured and ready for detecting surfaces 5 of materials 4, which it has learned.

During a production the learning process will usually be executed on a regular basis to ensure the safety of the program, adding new reference constitution or/and adding new criteria for grouping of already known sets of part values.

In another example (not shown) the material, which surfaces are to be detected can be a board of wood and the piece of wood will be sorted into three different grades, first quality wood, second quality wood and rejected wood.

The criteria for first quality wood is that there are no knots in the board of wood, the criteria for second quality wood is that there are no more than five knots on a board and that if there are between two and five knots none of these are placed less than twenty centimetres from each other. If there are more than five knots or if the two knots are less than twenty centimetres from each other the board of wood will be rejected.

In a newly installed apparatus wherein a method according to the invention is implemented the apparatus must first be configured. This is done by calibrating the detecting device, which is done in a conventional way already known in the art and depending on the type of detecting device used. Afterwards the apparatus is 'taught' how a knot looks like by feeding it a number of boards whereon different types of knots are marked by circling the knot with a coloured marker.

When the apparatus has learned and stored reference values representing the knots, criteria are set up for the size of the knots, their acceptable distance from each others and the acceptable numbers of knots on a board.

Based on the input from detecting the marked boards and the criteria set up by the user the method according to the invention is able to define the subset R and thereby avoiding computing other representative values which are not necessary for detecting the knot and also define the size of a part of the detected surface so that it will detect and process knots which are of a relevant size and ignore knots which are below the set criteria for the size and therefore irrelevant for the further processing.

Now the apparatus is configured and ready to detect unknown boards, which is done as described above with reference to fig.3.

In an advantageous embodiment the user is able to manually adjust different values for selectively adjusting the process.

The apparatus further can be configured to request a user to categorise an unknown constitution into a known constitution when the apparatus 1 may detect a constitution 13 that have representative values 20 which are not recognised to represent a known constitution. Thus the method described is typical used for sorting and/or scrapping materials such as wood, metal, plastic, cloth, paper, and/or such materials that are painted and/or printed with a picture and/or an inscription.

It shall be understood that the above described and on the drawing shown embodiments only are exemplifications and that many other embodiments are within the scope of the invention.

Claims

Claims
1. A method for classifying the quality of a surface (5) of a material (4) in relation to at least one reference constitution, comprising the steps of storing in a computer reference values (27) representing the at least one reference constitution, detecting the constitution (13) of the surface (5), - reading in the results of the detection as input to the computer, converting the input to sets of. part values (23) representing each a part of the surface (24) , comparing said values with the reference values (27) stored in the computer, reading out the comparisons as output from the computer, and classifying the surface quality of the material (4) in accordance with said output.
2. A method according to claim 1, characterised in by means of a program of the computer to calculate, values for the deviation of the sets of part values (23) from reference values (27) , representing at least one first reference surface constitution, values for the grading of the part values into classes in accordance with predetermined criteria for the deviations, values for a number of total values, which are representing each a group (35) of a predetermined minimum number of part values for the constitution of abutting and/or closely spaced surface parts (24) having part values in the same class, values for the deviation of said total values from reference values representing at least one second reference surface constitution, values for the grading of the total values into classes in accordance with other predetermined criteria for the deviations, whereby, - at least some of the calculated values are read out as output from the computer, and the surface quality of the material (4) is classified in accordance with said output.
3. A Method according to claim 1, 2 or 3, characterised in, that the results of the detecting of the constitution (13) of the surface (5) are read into the computer in form of digital signals, representing each the minimum surface area (17) , which can be detected with a detecting device (8) having a given resolution, that the digital signals are converted to values (20) representing each one of said minimum areas, assembling a number of said values (20) to sets of part values (23) representing each a part of the surface (24), - that at least some of the calculated values are converted to digital signals, and that said digital signals are read out as output from the computer .
4. A method according to claims 1, 2, 3 or 4, characterised in that the surface parts (24) are classified by predetermined criteria for deviations and that the groups (35) are classified by predetermined criteria for number of groups per unit of surface to be detected.
5. A method according to claims 1 - 4, characterised in that the groups (35) are arranged in a matrix which represents the surface mapping into the groups, and that the matrix is read out as signals representing the positioning of said groups on the surface (5) .
6. A method according to any of the preceding claims, characterised in that the constitution (13) of the surface (5) is determined by a statistical classification of the total values .
7. A method according to claim 6, characterised in that, the statistic classification is performed by the formula,
PC(Z) = χ,μ e Rn ∑ e R"x" n,c N (32)
Figure imgf000025_0001
where χ is the first values, • is the covariant matrix and • is the mean vector for the group c.
8. A method according to any of the preceding claims, characterised in comprising the steps, - selecting a material (4) with a surface constitution (13), which corresponds to the at least one reference constitution, detecting the surface constitution of the material, reading into the computer the detected result in form of digital signals, converting the digital signals to reference values (27), and storing the reference values .
9. A method according to any of the preceding claims, characterised in that the first values representing a constitution on the detected surfaces are selected by the following algorithm: Let D be the number of original first values . Let 0 be the amount of original first values; 0 =
{Fi|i=l,...,D}. Let S(Fi;Fj) denote the degree of difference between first values Fi and Fj , where the degree of difference is the maximum information compression index λ2 : λ ιa,j) = var( + var ") - sqrt((var(i) + var(y))2 -4 var(t)var( )(l - p(i, j)2)) . Let r denote the degree of difference between first values Fi and its k closest neighbour in R, where R is a subset of 0, which is derived by the steps: i) Chose a starting value for, 0<k<=D-l. Let the starting value for R is 0. j) Calculate rki for each first values Fi belonging to R. k) Find a first value F'i, where ri' is minimal. Keep F'i in R and remove the k closest first value F'i from R. Let ε be rk ' . 1) If k>|R|-l:k=|R|-l. m) If k=l:go to h.
Figure imgf000026_0001
iv) k=k-l. rki'=infFi belongs RΛ' v) If k=l:go to h. vi) Go to 6. o) Go to b. p) Pick R to be the reduced number of first values.
10. A method according to any of the preceding claims, characterised in that, the output read out from the computer is a graphical representation on a display or a printout on a sheet of paper with indication of at least the positioning of the groups (35) and the classes of the constitution (13) of surface parts (24) of which the groups are consisting.
11. A method according to any of the preceding claims, characterised in that the output is in shape of signals representing the classes of the groups and the surface parts (24) of these.
12. A method according to any of the preceding claims, characterised in that, the detection of the surface is carried out by means of laser scanning, ultrasound scanning or photography by for example an x-ray camera, a CCD-ca era or a CMOS-earnera.
13. A method for sorting a material in dependence of the quality of the surface of the material, characterised in that the quality of the surface is determined by using the method according to claims 1 - 12 and that the material is sorted in accordance with the classes read out as output from the computer.
14. A method for sorting a material accordance to claim 13 , characterised in that the method is used for sorting and/or scrapping materials such as wood, metal, plastic, cloth, paper, and/or such materials that are painted and/or printed with a picture and/or an inscription.
PCT/DK2003/000923 2002-12-19 2003-12-19 A method for classifying the quality of a surface WO2004057317A1 (en)

Priority Applications (2)

Application Number Priority Date Filing Date Title
DKPA200201954 2002-12-19
DKPA200201954 2002-12-19

Publications (1)

Publication Number Publication Date
WO2004057317A1 true true WO2004057317A1 (en) 2004-07-08

Family

ID=32668615

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/DK2003/000923 WO2004057317A1 (en) 2002-12-19 2003-12-19 A method for classifying the quality of a surface

Country Status (1)

Country Link
WO (1) WO2004057317A1 (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
DE102010047499A1 (en) 2010-04-12 2011-10-13 Fernando Puente León Automatic system for visually inspecting surface areas, has adjustable component that is provided with electronic image capture apparatus and illumination unit so that image of same zone is picked-up at different incident angles
US9239296B2 (en) 2014-03-18 2016-01-19 Corning Incorporated Skinning of ceramic honeycomb bodies

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5274714A (en) * 1990-06-04 1993-12-28 Neuristics, Inc. Method and apparatus for determining and organizing feature vectors for neural network recognition
DE19637234A1 (en) * 1996-09-13 1998-03-26 Michael F Braun Reducing optical definition for colour measurements of moving, textured, patterned objects for quality control
US6122065A (en) * 1996-08-12 2000-09-19 Centre De Recherche Industrielle Du Quebec Apparatus and method for detecting surface defects
DE19940879A1 (en) * 1999-08-27 2001-03-08 Innomess Elektronik Gmbh Device and procedure for comparison of a digitized print image with a reference image for automatic quality control so that if error values exceed a threshold value an alarm is generated to inform print machine operators
EP1092973A1 (en) * 1999-10-12 2001-04-18 Perceptron, Inc. A system and method for detecting defects in a surface of a workpiece
US20020051564A1 (en) * 1998-12-04 2002-05-02 Norbert Benesch Method and device for optically monitoring fabrication processes of finely structured surfaces in a semiconductor production
WO2002060605A1 (en) * 2001-01-30 2002-08-08 E.L.C.O.S. S.R.L. Method of classifying the aesthetic quality of planar materials, and system for automatically sorting planar materials using such a method

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5274714A (en) * 1990-06-04 1993-12-28 Neuristics, Inc. Method and apparatus for determining and organizing feature vectors for neural network recognition
US6122065A (en) * 1996-08-12 2000-09-19 Centre De Recherche Industrielle Du Quebec Apparatus and method for detecting surface defects
DE19637234A1 (en) * 1996-09-13 1998-03-26 Michael F Braun Reducing optical definition for colour measurements of moving, textured, patterned objects for quality control
US20020051564A1 (en) * 1998-12-04 2002-05-02 Norbert Benesch Method and device for optically monitoring fabrication processes of finely structured surfaces in a semiconductor production
DE19940879A1 (en) * 1999-08-27 2001-03-08 Innomess Elektronik Gmbh Device and procedure for comparison of a digitized print image with a reference image for automatic quality control so that if error values exceed a threshold value an alarm is generated to inform print machine operators
EP1092973A1 (en) * 1999-10-12 2001-04-18 Perceptron, Inc. A system and method for detecting defects in a surface of a workpiece
WO2002060605A1 (en) * 2001-01-30 2002-08-08 E.L.C.O.S. S.R.L. Method of classifying the aesthetic quality of planar materials, and system for automatically sorting planar materials using such a method

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
DE102010047499A1 (en) 2010-04-12 2011-10-13 Fernando Puente León Automatic system for visually inspecting surface areas, has adjustable component that is provided with electronic image capture apparatus and illumination unit so that image of same zone is picked-up at different incident angles
ES2381723A1 (en) * 2010-04-12 2012-05-31 Asociacion De Investigacion Y Desarrollo En La Industria Del Mueble Y Afines (Aidima) Artificial vision system for detecting defects in finished surfaces
US9239296B2 (en) 2014-03-18 2016-01-19 Corning Incorporated Skinning of ceramic honeycomb bodies

Similar Documents

Publication Publication Date Title
US7515772B2 (en) Document registration and skew detection system
US5194720A (en) Method and apparatus for performing on-line integrated decoding and evaluation of bar code data
US5431288A (en) Mail sorting apparatus
US7209599B2 (en) System and method for scanned image bleedthrough processing
US4567610A (en) Method of and apparatus for pattern recognition
US4516264A (en) Apparatus and process for scanning and analyzing mail information
US6166393A (en) Method and apparatus for automatic inspection of moving surfaces
US5809165A (en) Method for color control in the production process
US5118195A (en) Area scan camera system for detecting streaks and scratches
US5257101A (en) Apparatus and method for selecting a log orientation and pattern for sawing
US5703960A (en) Lumber defect scanning including multi-dimensional pattern recognition
Zheng et al. Automatic inspection of metallic surface defects using genetic algorithms
US6291829B1 (en) Identification of recording medium in a printer
US5235652A (en) Qualification system for printed images
US6092059A (en) Automatic classifier for real time inspection and classification
Silvén et al. Wood inspection with non-supervised clustering
US5687250A (en) Image quality analysis method and apparatus
US7187995B2 (en) Maximization of yield for web-based articles
Tatzer et al. Industrial application for inline material sorting using hyperspectral imaging in the NIR range
US6975949B2 (en) Full width array scanning spectrophotometer
US5652804A (en) Automatic inspection of printing plates or cylinders
US6005683A (en) Document edge detection by linear image sensor
US5617484A (en) Image binarizing apparatus
US5970166A (en) System or method for identifying contents of a semi-opaque envelope
US5481619A (en) Inspection method using templates images, unique histogram analysis, and multiple gray levels correlated to addresses of template images

Legal Events

Date Code Title Description
AK Designated states

Kind code of ref document: A1

Designated state(s): AE AG AL AM AT AU AZ BA BB BG BR BY BZ CA CH CN CO CR CU CZ DE DK DM DZ EC EE EG ES FI GB GD GE GH GM HR HU ID IL IN IS JP KE KG KP KR KZ LC LK LR LS LT LU LV MA MD MG MK MN MW MX MZ NI NO NZ OM PG PH PL PT RO RU SC SD SE SG SK SL SY TJ TM TN TR TT TZ UA UG US UZ VC VN YU ZA ZM ZW

AL Designated countries for regional patents

Kind code of ref document: A1

Designated state(s): GH GM KE LS MW MZ SD SL SZ TZ UG ZM ZW AM AZ BY KG KZ MD RU TJ TM AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HU IE IT LU MC NL PT RO SE SI SK TR BF BJ CF CG CI CM GA GN GQ GW ML MR NE SN TD TG

121 Ep: the epo has been informed by wipo that ep was designated in this application
122 Ep: pct application non-entry in european phase
NENP Non-entry into the national phase in:

Ref country code: JP

WWW Wipo information: withdrawn in national office

Country of ref document: JP