materials

Article
Automated Control of Surface Defects on Ceramic
Tiles Using 3D Image Analysis
Andrzej Sioma
Faculty of Mechanical Engineering and Robotics, Department of Process Control, AGH University of Science
and Technology, 30-059 Krakow, Poland; [email protected]; Tel.: +48-12-6175005




Received: 11 February 2020; Accepted: 7 March 2020; Published: 10 March 2020


Abstract: This paper presents a method of acquisition and analysis of three-dimensional images in
the task of automatic location and evaluation of defects on the surface of ceramic tiles. It presents a
brief description of selected defects appearing on the surface of tiles, along with the analysis of their
formation. The paper includes the presentation of the method of constructing a 3D image of the tile
surface using the Laser Triangulation Method (LTM), along with the surface imaging parameters
employed in the research. The algorithms of three-dimensional surface image analysis of ceramic tiles
used in the process of image filtering and defect identification are presented. For selected defects, the
method of measuring defect parameters and the method of visualization of defects on the surface are
also presented. The developed method was tested on defective products to confirm its effectiveness
in the field of quick defect detection in automated control systems installed on production lines.

Keywords: surface defects of ceramic tiles; 3D vision system; 3D image; 3D image analysis

1. Introduction
The increase in the capacity of production lines producing ceramic tiles imposes the introduction
of innovative solutions in the field of fault diagnosis and detection systems on product surfaces.
In many manufacturing enterprises, the quality inspection of ceramic tiles is carried out by a person
supported by sensor or vision systems. Firat et al. presented the study of the description of employee
fatigue and the possibility of errors made by them in the implementation of visual assessment of the
product related to work time or type of product [1].
Introduction of vision systems to the production lines made it possible to acquire an image and
analyse this image in order to identify and locate defects. As part of the vision methods described in
the literature, the analyses of shades of grey or colours are usually used. Hanzaeia et al. presented a
system based on 2D image analysis with statistical methods to detect the edges of such surface defects
as cracks and scratches [2]. Hocenski et al. presented a computer system (CVS) that enables the analysis
of the image of tiles and the assessment of defects on their surface using a program written in C ++
with OpenCV libraries used to analyse 2D images [3]. Job Samarawickrama et al. presented a system
based on image analysis implemented in the Matlab software that also uses two-dimensional image
analysis, both in monochrome and colour [4]. Vaideliene et al. proposed the use of wavelet analysis for
the location of defect areas [5]. The authors presented an effective defect detection system on texture
images visible on the surface of tiles. In [6], the authors described a methodology for automatically
identifying granite varieties. They used spectral information captured by a spectrophotometer and
machine learning techniques.
In article [7], the authors presented hybrid 3D-2D laser scanning system for the structural defects
recognition of slate slabs. The paper describes defects in roofing slates, parameters describing their
quality and method of those parameter control. In [8,9], the authors present machine learning methods

Materials 2020, 13, 1250; doi:10.3390/ma13051250 www.mdpi.com/journal/materials

Materials 2020, 13, 1250 2 of 13

in the task of identifying defective slate slabs and present the differences in the results to the adopted
learning method.
In most industrial applications, the image analysis system is responsible for detecting defects on
a surface. The task of checking the geometry of the surface in the form of a flatness check is done
by means of a sensor system (4–8 sensors); these measure the height in designated areas of the tile
surface. Most vision systems used in the ceramic industry can be described as two-dimensional due
to the information available in the image and the method of surface imaging. They are most often
implemented as additional equipment in technological machines and checking stations, the task of
which is the implementation of inter-operational inspection of product parameters. Introduction of
automated video stands with extended measuring and inspection capabilities is required, especially
in mass production carried out in the “zero defects” system, providing a thorough inspection of all
products leaving the production line.
This paper presents the method of acquisition and analysis of 3D images in the inspection tasks of
selected parameters of a ceramic tile. The paper presents examples of using 3D images in the implementation
of tasks to check defects in geometry and surface defects of the product. The advantage of the method is
that the three-dimensional image enables the implementation of a much larger number of measurements
and analyses, especially in the field of spatial measurements, complementing and extending the currently
used vision methods in the video industry based on two-dimensional image analysis [10–15].
The geometric defects occurring most often on the surface of ceramic tiles include defects in the
flatness of the tile’s usable surface. These defects are visible in the form of curved corners or edges or
in the form of a visible buckling on the surface. Geometric defects also include any damages to edges
and corners. A large group of defects are also flaws of usable surfaces visible in the form of surface
delamination, scratches or local roughness variations. These defects may be very small in size or may
be seen on almost the entire surface of the tile. The following drawing shows a large delamination
area, located in the centre of the ceramic tile. Implementation of research and development of a system
enabling detection of all the described defects requires implementation of imaging of the entire surface
of the tile and preparation of a set of reference algorithms to enable the detection of each type of defects.
The objective of the research described in this paper was to develop and implement a system that
allows comprehensive assessment of the quality of ceramic tiles. It results from the need to check
the quality of the product’s performance in the scope of checking the geometrical dimensions of the
ceramic tile combined with simultaneous checking of the presence and parameters of defects on the
surface of ceramic products. For this purpose, a vision system was developed using the LTM method,
designed to build a three-dimensional image of the surface of the tile. It was assumed that the video
system operation parameters and 3D image analysis algorithms will enable location, description of
defect parameters and a comprehensive assessment of the ceramic tile geometry.

2. Materials and Methods

Three-dimensional vision systems are configured and built strictly with a view of the implementation
of inspection and measurement tasks. This requires a detailed analysis of the task in terms of the possibility
of determining the parameters describing the tested product. It is necessary to carry out imaging tests
both in laboratory conditions and in the conditions of the production line. The 3D vision system presented
in this paper was designed and launched by a research team committed to the automation of the product
performance inspection process and by specialists involved in the technology of ceramic tile production.
Due to the method of three-dimensional image construction being used in the paper, and using the laser
triangulation method as part of the preliminary research, the analysis of the cooperation of the image
recording system, laser lighting and the examined surface was performed as a function of the adopted
configuration of the vision system.
The first stage of the work was to prepare a catalogue of defects found in the products. The task
included a description of the defects and characteristic parameters describing each of the defects. In the
next stage, a technical solution was developed, enabling three-dimensional imaging of the product
 Materials 2019, 12, x FOR PEER REVIEW 3 of 13
Materials 2020, 13, 1250 3 of 13
The first stage of the work was to prepare a catalogue of defects found in the products. The task
included a description of the defects and characteristic parameters describing each of the defects. In
the nextand
surface stage, a technical
defects solution catalogue
for the adopted was developed, enabling
of defects. three-dimensional
Next, imaging
three-dimensional imageofanalysis
the
product surface and defects for the adopted catalogue of defects. Next, three-dimensional
algorithms were developed allowing for a parametric evaluation of each of the defects andimage
marking
analysis
the defectalgorithms wereof
on the surface developed allowing for a parametric evaluation of each of the defects and
the product.
marking the defect on the surface of the product.
2.1. Characteristics of the Material for Testing
2.1. Characteristics
Ceramic tiles of the three
with Material for Testing
types of production defects were used for the tests. It was assumed that
Ceramic
the product tiles with
analysis of three types oftile
the ceramic production defects
surface will were used forinthe
be implemented tests.
the scopeIt was assumed that
of checking:
the product analysis of the ceramic tile surface will be implemented in the scope of checking:
• Microcracking: i.e., cracks visible on irregularly shaped surface, several centimetres long and 100 to
• 200Microcracking:
µm wide. The i.e.,morphology
cracks visible
of on irregularlywas
microcracks shaped surface, several
characterized centimetres of
by the occurrence long and
numerous
100 to 200 μm wide. The morphology of microcracks was characterized by the occurrence
bifurcations in the form of scratches and delaminations visible on the surface of the tiles. Shortcomings of
numerous
of bifurcations
such morphology in most
arise the form
oftenofas
scratches
a result and delaminations
of thermal stresses visible on the surface
of the second ofappear
kind that the
tiles. Shortcomings of
during the firing process. such morphology arise most often as a result of thermal stresses of the
second kind that appear during the firing process.
• Tile thickness deformation: with ±1000 µm deviation value, visible in the form of buckling or
• Tile thickness deformation: with ±1000 μm deviation value, visible in the form of buckling or
defects on the surface. This type of production defect is often associated with non-homogeneous
defects on the surface. This type of production defect is often associated with non-homogeneous
distribution of granules filled into the mould before the pressing process.
distribution of granules filled into the mould before the pressing process.
• Geometrical defects: characterized by non-parallelism of edges and aberration of angles of tile
• Geometrical defects: characterized by non-parallelism of edges and aberration of angles of tile
corners
cornerswith
withaadeviation
deviationofofupuptoto
several
severaldegrees
degreesfrom thethe
from right angle.
right ThisThis
angle. typetype
of defect is related
of defect is
to the occurrence of uneven shrinkage of the tile occurring during the sintering
related to the occurrence of uneven shrinkage of the tile occurring during the sintering process. process.

Figure 1 presents
Figure presentssample
sampleimages
imagesofof typical production
typical production defects thatthat
defects are most frequently
are most foundfound
frequently in
the
in theproduction process.
production TheThe
process. above defects
above are characteristic
defects of ceramic
are characteristic tile production
of ceramic processes
tile production and
processes
theirtheir
and recognition and and
recognition limitation in theinproduction
limitation process
the production are theare
process basic
the tasks
basicfor thefor
tasks producers of tile of
the producers
ceramics.
tile ceramics.

Figure 1.
Figure 1. Images
Images ofof typical
typical production
productiondefects
defectsofoftiles
tilessampled
sampledforforanalysis: (a)(a)
analysis: microcracks with
microcracks with
irregular shape,
irregular shape, several
several centimetres
centimetreslong,
long,100100toto200
200μmµmwide,
wide, (b)(b)
deformation
deformationof of
tiletile
thickness with
thickness with
defects on
defects on the
the surface
surface of
of the
the ceramic
ceramictile,
tile,(c)
(c)edge
edgedefects
defectsininthe
theform
formofof
material chipping.
material chipping.

2.2. Equipment
2.2. Equipment
Figure22presents
Figure presentsa amodel
modelof of a three-dimensional
a three-dimensional vision
vision system
system for detecting
for detecting and and assessing
assessing surface
surface defects of tiles using the LTM method [16]. In order to assess production defects of ceramic
defects of tiles using the LTM method [16]. In order to assess production defects of ceramic tiles,
tiles,
the the measurement
measurement systemsystem presented
presented in Figurein 2a
Figure
was 2a waswhich
used, used,includes:
which includes: 650 nm
650 nm laser laser
illuminator
illuminator
with a power with a power
of 40 of 40 mW with
mW equipped equipped with ansystem
an optical opticalto system to the
control control the of
width width of theline
the laser laserand
line and an additional control system enabling control of the laser power. The
an additional control system enabling control of the laser power. The Z-Laser type ZM18 laser was Z-Laser type ZM18
laser was
used. used. was
The work The used
work awas usedE55
Ranger a Ranger
cameraE55 camera
(SICK AG, (SICK AG, Waldkirch,
Waldkirch, Germany) withGermany)
a matrixwith a a
with
matrix with
resolution ofa1536
resolution of 1536[16].
× 512 pixels × 512 pixels [16].
Thevision
The visionsystem
systemworks
worksonon a control
a control stand
stand designed
designed for needs
for the the needs
of theofmeasuring
the measuring task.
task. Ceramic
Ceramic tiles are placed on a platform that moves relative to the permanently mounted
tiles are placed on a platform that moves relative to the permanently mounted laser–camera system. laser–camera
The displacement is carried out in the Y-axis of the coordinate system associated with the 3D image.
The station propulsion system uses a stepper motor with a propeller with a pitch of 2 mm. The system
Materials 2019, 12, x FOR PEER REVIEW 4 of 13
Materials 2020, 13, 1250 4 of 13
system. The displacement is carried out in the Y-axis of the coordinate system associated with the 3D
image. The station propulsion system uses a stepper motor with a propeller with a pitch of 2 mm.
The system isfrom
is controlled controlled from aenabling
a controller controller
theenabling the implementation
implementation of a minimum of adisplacement
minimum displacement
of 0.01 mm.
of 0.01 mm. Such displacement is additionally controlled by two encoders. The first is built in axis
Such displacement is additionally controlled by two encoders. The first is built in the engine the
engine
and theaxis andisthe
second secondon
mounted is the
mounted on the
propeller. Thispropeller. This enablescontrol
enables displacement displacement control and
and identification of
identification of platform positioning errors.
platform positioning errors.
The
The geometry
geometryof of the
the 3D
3D vision
vision system
systemhashas been
beenconfigured
configuredforfor the
the implementation
implementationof ofthe
thetask,
task,
including
includingthe thefollowing
followingparameters:
parameters:dimensions
dimensions of of
thethe
examined
examined “tile” object
“tile” 192 192
object mmmm × 192× mm
192 and
mm
the required imaging resolution of 0.25 mm in the X- and Y-axes of the coordinate system of the
and the required imaging resolution of 0.25 mm in the X- and Y-axes of the coordinate system of the
2
measurement
measurementstand.stand. This
Thissystem
systemallows
allowsimaging
imagingof of11mm
mm2 atataaresolution
resolutionof of0.25
0.25mm
mmon on the
the plane
plane
with 16 measurement points.
with 16 measurement points.

Figure 2.
Figure View of
2. View of the
the workplace
workplace for
for the
the detection
detection and
and assessment
assessment of of production
production defects
defects of
of ceramic
ceramic
tiles: (a) the mounting of the camera and laser illuminator relative to the ceramic tile
tiles: (a) the mounting of the camera and laser illuminator relative to the ceramic tile surface, (b) surface, (b) the
the
layoutof
layout ofthe
thepoints
pointsdescribing
describingthe
theheight
heightprofile
profile on
on the
the tile
tilesurface
surface in
in the
theplane
planeof
ofillumination
illuminationforforthe
the
adopted configuration of the 3D vision system.
adopted configuration of the 3D vision system.
2.3. Measurement Methodology
2.3. Measurement Methodology
The image of the three-dimensional surface of the tile is constructed using surface imaging
The image
illuminated withofa the three-dimensional
linear laser beam presented surface of the 2a.
in Figure tile Along
is constructed
the laser using surfacemeasuring
line, further imaging
illuminated with a linear laser beam presented in Figure 2a. Along the laser line,
points are determined enabling the tile height to be described in each of the points (Figure 2b). These further measuring
points
points are determined
together form a enabling the tile
height profile height tothe
describing beshape
described
of theinsurface
each ofonthethe
points (FigureFor
X–Z plane. 2b). These
surfaces
points together form a height profile describing the shape of the surface on
without defects, it will be a straight line parallel to the X-axis of the coordinate system. For damaged the X–Z plane. For
surfaces without
areas, it will be a defects, it will be the
curve describing a straight
shape of line
theparallel
surfacetodefect.
the X-axis of the coordinate system. For
damaged areas, it will be a curve describing the shape
Three-D imaging requires determining a set of height profiles of the surfacethatdefect.
are evenly distributed on the
Three-D imaging requires determining a set of height profiles that are evenly distributed on the
surface of the tile. This is related to the tile displacement relative to the vision system in the direction
surface of the tile.
of the Y-axis. The This is related of
displacement to the
the test
tile displacement
object relativerelative to thesystem
to the vision vision system in the direction
in the determination of
of the Y-axis. The displacement of the test object relative to the vision system in the determination of
successive height profiles is determined as the measurement resolution in the Y-axis of the measurement
successive
position. The height profiles
assembly of is determined
a set as the measurement
of height profiles thus determined resolution
that are in the Y-axis
spread evenly ofon the
the
measurement position. The assembly of a set of height profiles thus determined that are spread
surface of the tile enables the construction of a three-dimensional image of the surface of the ceramic
evenly on the surface of the tile enables the construction of a three-dimensional image of the surface
tile. To calibrate the camera geometry, a built-in model for Ranger cameras was used. Calibration
of the ceramic tile. To calibrate the camera geometry, a built-in model for Ranger cameras was used.
of the power of the laser illuminator and the frequency of the camera’s operation were selected on
Calibration of the power of the laser illuminator and the frequency of the camera’s operation were
the basis of statistical tests carried out on prepared sets of tiles manufactured correctly and tiles with
selected
damagedonsurfaces
the basis fromof one
statistical tests carried
production out ontests
line. Imaging prepared sets of tilesfor
were performed manufactured
twelve differentcorrectly
plates
and tiles with damaged surfaces from one production line. Imaging tests were
for one plate dimension. Calibration of lighting and system operating frequency should be performed performed for twelve
different
separatelyplates for product
for each one plate dimension.
group Calibration
due to changes in theof lighting
surface and system
structure operating
of the tiles frequency
and their colours.
should be performed separately for each product group due to changes in the surface structure of the
tiles and their colours.
Materials 2020, 13, 1250 5 of 13

2.4. Setting the Height Profile and Determination of Measurement Resolution

In the study of the surface of the ceramic tile, the geometry shown in Figure 3a was used, in
which the sensor of the vision system is placed at a 90-α angle to the surface of the measuring table.
The laser beam is perpendicular to the tested surface and, at the same time, set at the angle α to the
optical axis of the camera (Figure 3a). The height profile was determined based on the image analysis
of the laser line projected on the surface, which is the equivalent of the cross-section in the adopted
laser position. The image of the laser line is recorded by the vision system set at the angle of 90-α in
the manner presented in Figure 3b. The image recorded by the camera is subjected to analysis that
allows determination of a set of points forming the surface height profile. The points that describe
the height profile are determined with the resolution that depends on the adopted configuration, the
optical system used and the resolution of the matrix of the vision system.
Determination of the resolution of the 3D vision system involves setting three components of the
resolution for each axis of the coordinate system associated with the control station. For the Z-axis,
this requires determining the minimum height variation of the object described by the ∆Z parameter,
at which the laser image is shifted by exactly one row of pixels on the matrix of the vision system
(Figure 3a). On the plane parallel to the plane of the matrix, the resolution ∆X in the X axis was
determined on the basis of the field of view dimensions (FOV: Field of View) and the resolution of the
matrix in pixels. The resolution ∆Y in the Y-axis is defined as the displacement of the tested object
when determining successive height profiles. Assuming the use of the 768 × 512 pixel matrix area and
choosing a video system lens for observation of a 192 mm wide object (FOV = 192 mm), the resolution
in the X and Z-axis was determined for the adopted configuration.

∆X = 192 mm/768 pixels = 0.25 [mm/pixel] (1)

When calculating the resolution in the Z-axis of the station, the approximation was used assuming
that the angle α is equal to the angle α1. In fact, the angle α1 is α: γ. Using this simplification, the
resolution in the Z-axis is determined for angle α = 45◦ .

∆Z = ∆X/sin(α) = 0.25/sin(45◦ ) = 0.35 [mm/pixel] (2)

where: ∆Z: the resolution in the Z-axis, ∆X: the resolution in the X axis, α: the angle between the
optical axis of the laser and the optical axis of the camera.
The resolution in the Y-axis of the coordinate system depends on the shift of the examined area
between successive image acquisitions by the vision system. Taking into account the determined
resolution specified in the X axis equal to 0.25 mm/pixel, the following values were assumed in the
Y-axis:
∆Y = 0.25 mm/pixel (3)

The configuration of the station with the 0.25 mm/pixel resolution in both the X axis and the Y-axis
enables description of each square millimetre of the area with 16 measurement points.
An analysis of the impact of camera angle relative to the laser plane on the resolution possible on
a 3D image was performed. The analysis was performed with the same matrix size of 1536 pixels for
three FOVs equal to 200 mm, 300 mm and 400 mm. The results of the calculations are presented on the
characteristics shown in Figure 3c.
Materials 2020, 13, 1250 6 of 13
Materials 2019, 12, x FOR PEER REVIEW 6 of 13

Figure 3. Configuration of the 3D vision system: (a) the configuration of the measurement system,
Figure 3. Configuration of the 3D vision system: (a) the configuration of the measurement system, (b)
(b) 2D image of the laser beam illuminating the surface of the tile registered by the camera, (c) resolution
2D image of the laser beam illuminating the surface of the tile registered by the camera, (c) resolution
graph depending for camera angle set for three selected fields of view(FOV).
graph depending for camera angle set for three selected fields of view(FOV).
2.5. 3D Image Construction
2.5. 3D Image Construction
A 3D image built using a 3D vision system is described by a matrix with dimensions (i, j).
A 3D imagej corresponds
The dimension built using a to 3Dthevision system is in
x coordinate described
the imageby acoordinate
matrix with dimensions
system (i, j). The
and defines the
dimension j corresponds to the x coordinate in the image coordinate system
width of the field of vision of the video system by specifying the number of columns of the matrix. and defines the width of
the field of vision of the video system by specifying the number of columns
The dimension i corresponds to the coordinate y in the coordinate system related to the image and of the matrix. The
dimension
defines the inumber
corresponds to the
of height coordinate
profiles y in the
determined forcoordinate
the tested system
object byrelated to thethe
specifying image and defines
number of rows
the number of height profiles determined for the tested object by specifying
in the matrix. The individual cells of the matrix contain information about the height h (x,y) of the the number of rows in
the matrix. The individual cells of the matrix contain information about the height
image determined with the resolution ∆Z for each of the points defining the surface of the ceramic tile. h (x,y) of the image
determined with the
Composition of resolution ΔZ forthe
profiles enables each of the points
construction ofdefining the surface ofimage
a three-dimensional the ceramic tile. the
describing
Composition of profiles enables the construction of a three-dimensional
surface with a resolution of ∆X = ∆Y = 0.25 mm in the X and Y-axis and with a resolution ∆Z = 0.35 mm image describing the
surface with a resolution of ΔX = ΔY = 0.25 mm in the X and Y-axis and with
in the Z-axis. Sample images of ceramic tile surfaces made with the adopted resolution are shown in a resolution ΔZ = 0.35
mm in the
Figure Z-axis.
4b,c. TheySample images ofwith
were produced ceramic tile source
a laser surfacesofmade20 mW.with theevery
For adoptedsquareresolution are shown
millimetre of the
in Figure 4b,c. They were produced with a laser source of 20 mW. For every
ceramic tile surface there are 16 measuring points allowing observation of the change in height square millimetre of the
in
ceramic tile
selected areas. surface there are 16 measuring points allowing observation of the change in height in
selected
Theareas.
use of 3D imaging and 3D image analysis of the ceramic tile surface may be presented in
the form use
The of anof 3D imaging the
algorithm, anddiagram
3D imageofanalysis
which of is the ceramic4a.
in Figure tileThe
surface may beispresented
3D image recordedinwith the
form of an
a vision algorithm,
system, the diagram
and then undergoes of which is in Figure
pre-processing, 4a.purpose
the The 3D ofimage
which is recorded withitafor
is to prepare vision
the
system, and then undergoes pre-processing, the purpose of which
implementation of the measurement procedures. The implementation of the measurement parameters is to prepare it for the
implementation of the measurement procedures. The implementation
describing the defects makes it possible to locate the fault area and quality classify the product. of the measurement
parameters describing the defects makes it possible to locate the fault area and quality classify the
product.
 Materials 2020, 13, 1250 7 of 13
Materials 2019, 12, x FOR PEER REVIEW 7 of 13

Figure 4. Schematic
Figure diagram
4. Schematic of the
diagram reconstruction
of the reconstruction measurement
measurementalgorithm,
algorithm,(a)
(a)algorithm
algorithm steps,
steps, (b)
image
(b) of the laser
image of theline
laseron ceramic
line surface,
on ceramic (c) 3D
surface, image
(c) 3D of surface.
image of surface.

Analysis of three-dimensional images is based on the processing of data describing the height
Analysis of three-dimensional images is based on the processing of data describing the height
of points forming the surface collected in a two-dimensional matrix. The purpose of this analysis is
of points forming the surface collected in a two-dimensional matrix. The purpose of this analysis is
to determine the characteristic features of the image, e.g., describing surface defects and presenting
to determine
them in thethe formcharacteristic
of parameters. features of the
The most image,
common e.g., describing
parameters surface defects
are: geometrical and presenting
dimensions, surface
them in the form of parameters. The most common parameters are: geometrical
areas, volume or shape of defects as defined in the catalogue of defects. Analysis of the image dimensions, surface
areas, volume
of the or shape of defects
three-dimensional surfaceasofdefined
ceramicin themay
tiles catalogue of defects.
be divided Analysis
into three stages:ofpre-processing,
the image of the
three-dimensional surface of ceramic tiles may be
measurements of selected parameters, evaluation of measurement results. divided into three stages: pre-processing,
measurements of selected parameters,
Three-dimensional images wereevaluation
pre-processed of measurement results. and improve the image
to remove distortions
Three-dimensional images were pre-processed to remove distortions
characteristics that will be used in the phase of measuring the parameters and improve
describing the[17].
the defects image
characteristics
Pre-processing thatiswill be used
carried out in the point,
using phase context
of measuring the parameters
and morphological describing theThe
transformations. defects
point[17].
Pre-processing
transformations is carried out using
of the image point, out
are carried context
point and morphological
by point using the heighttransformations.
value of eachThe point
of the
transformations of the image are carried out point by point using the height value of each of the
points forming the surface. Three-dimesnional image context transformations allow to transform the
value
points of eachthe
forming point takingThree-dimesnional
surface. into account the value imageof the points
context in its vicinity. Inallow
transformations the pre-processing
to transform the
phase, morphological transformations are also carried out, allowing, for example, improvement of
value of each point taking into account the value of the points in its vicinity. In the pre-processing
such image features as edge visibility, edge and surface smoothing. These transformations should
phase, morphological transformations are also carried out, allowing, for example, improvement of
be implemented in case of objects with a complex structure and a large number of edges. In case of
suchceramic
image tiles,
features as edge visibility, edge and surface smoothing. These transformations should be
morphological transformations will be used to detect the edges of cracks and microcracks
implemented
on the surface.in case of objects
As part of the 3D with
imagea complex
filtering, astructure
3 × 3 median andfilter
a large
was number
used. This ofprotects
edges. against
In case of
ceramic
single random changes of values on individual pixels. At the same time, a small 3 × 3 mask does notand
tiles, morphological transformations will be used to detect the edges of cracks
microcracks
hide cracks onofthe surface.
small widths.As part of the
Additional 3D image
protection filtering,
against the lacka of3 ×information
3 median aboutfilter the
washeight
used.onThis
protects against
a single pixel single
was therandom
use of a changes of morphological
Dilate type values on individual
filter in 3pixels.
× 3 maskAt the same time,
for surface a small
analysis. For 3 ×
3 mask
crackdoes not hide
detection, the cracks of small
Erode type widths. Additional
of morphological filters wasprotection against artificial
used. It allowed the lack enlarging
of information
of
about the height
cracks on a single
in a one-pixel width.pixel was the use of a Dilate type morphological filter in 3 × 3 mask for
surface analysis. For crack detection, the Erode type of morphological filters was used. It allowed
3. Results
artificial enlarging of cracks in a one-pixel width.
Figure 5 shows a three-dimensional image of the surface with a flatness defect. The left part of the
3. Results
image shows the flat correct surface of the tile described as “Normal or undamaged surface”. On the
right side of the tile, there is a defect in the form of a bump. The parametric evaluation of this defect
Figure
was made5 on
shows a three-dimensional
the basis of the measurement image
of theofdeviation
the surface with the
between a flatness defect.ofThe
actual surface the left part of
tile and
the image showssurface
the reference the flatassumed
correct as
surface of theflat
a perfectly tilepattern.
described as “Normal or undamaged surface”. On
the right side of the tile, there is a defect in the form of a bump. The parametric evaluation of this
defect was made on the basis of the measurement of the deviation between the actual surface of the
tile and the reference surface assumed as a perfectly flat pattern.
Materials 2020, 13, 1250 8 of 13
Materials 2019, 12, x FOR PEER REVIEW 8 of 13
Materials 2019, 12, x FOR PEER REVIEW 8 of 13

Figure 5. AA three-dimensional
three-dimensional image
image of the surface of aa ceramic tile with
with aaa marked normal area and aa
Figure 5.
Figure 5. A three-dimensional image of the
of the surface
surface of
of a ceramic
ceramic tile
tile with marked normal
marked normal area
area and
and a
defect in the form of a bump.
defect in
defect in the
the form
form of
of aa bump.
bump.

Figure
Figure 666 presents
Figure presents
presentsaaacomparison
comparison
comparisonofof the
ofthe actual
theactual
actual and
and
and reference
reference
reference surfaces
in in
surfaces
surfaces in the
thethe area
area
area of
of the
of the the defect
defect
defect on
on
on the
the surface
the surface
surface of the
of tile
of the tile
the in in order
tileorder
in order to visualize
to visualize
to visualize and
and and display
display
display the deviation
the deviation
the deviation from
fromfrom the ideal
the ideal
the ideal surface.
surface.
surface.

Figure 6. A three-dimensional view of the reference surface describing the correct surface of the tile
Figure 6. A three-dimensional view of the reference surface describing the correct surface of the tile
Figure
and the6.actual
A three-dimensional
surface with the view
defect.of the reference surface describing the correct surface of the tile
and the actual surface with the defect.
and the actual surface with the defect.
The next step in the implementation of the planar surface flatness algorithm is to determine the
The next
Thearea
next step in in the implementation of the
the planar surface flatness algorithm is
is to
to determine the
surface ofstep the that
the defect implementation
is outside theoftolerance planar surface
value flatness
adopted by thealgorithm
quality control determine
department.the
surface
surface area
area to of the defect
of determine
the defectthe that
that is outside
is outside the tolerance
the intolerance value
value adopted by the
adopted by algorithm quality
the quality control
control
In order height deviation the image, a measurement is used to
department.
department.
determine the values describing the height of the surface along the defined measurement profiles set in
In
In order
order to determine
torows
determine the
the height
height deviation
deviation in
in the
the image, aa measurement algorithm is used to
the subsequent of the image matrix. It is a type of image,
scanning measurement algorithm
that allows locating areasiswith
usedtheto
determine
determine the values
the values describing
describing the height
the height ofof the surface along the defined measurement profiles set
highest flatness deviation and determining thethe surface
shape along
of the thevisible
defect defined onmeasurement
the surface. profiles set
in
in the subsequent
theFigure
subsequent rows
rows of
of the
the image
image matrix.
matrix. It
It is
is aa type
type of
of scanning
scanning that
that allows
allows locating
locating areas
areas with
with
7 presents a sample measurement profile showing both the value of flatness deviation and
the highest
the shape flatness
highestofflatness deviation
deviation and determining
and determining the shape
the shape of the
of the defect visible
defectisvisible on the surface.
the the defect on the surface of the ceramic tile. The analysis carriedon the
out insurface.
each row of the
Figure 77 presents aa sample measurement profile showing both the
the value
valuetoof
ofa flatness deviation
imageFigure
enablingpresents
detection sample
of a defectmeasurement
with a field profile
of 1 × 1showing
px, whichboth corresponds flatness
field deviation
of 0.25 mm ×
and
and the
the shape
shape of
of the
the defect
defect on
on the
the surface
surface of
of the
the ceramic
ceramic tile.
tile. The
The analysis
analysis is
is carried
carried out
out in
in each
each row
row
0.25 mm.
of the image
of theThe
image enabling
enabling detection
detection of a defect
of aoutside with
defect with a field of 1 × 1 px,
a field ofis1determined which
× 1 px, which corresponds to a
a field of
field
shape and area of the area the tolerance by corresponds
comparing the to height of
of
0.25
0.25 mm
mm ×× 0.25
0.25 mm.
mm.
each pixel that creates the actual surface image with the height value of the reference plane. It is a point
transformation allowing us to determine the surface area of the defect area. An example of the defect
area may be found in Figure 8. Image analysis also allows us to determine the edges of the defect,
the centre of mass of the defect and the perimeter of the area describing the defect. These features
are shown in the figure below as additional parameters describing the defect, and are used in the
quantitative assessment of the defect.
 Materials 2020, 13, 1250 9 of 13
Materials 2019, 12, x FOR PEER REVIEW 9 of 13

Figure 7. Measurement of the height deviation distribution in relation to the reference plane.

The shape and area of the area outside the tolerance is determined by comparing the height of
each pixel that creates the actual surface image with the height value of the reference plane. It is a
point transformation allowing us to determine the surface area of the defect area. An example of the
defect area may be found in Figure 8. Image analysis also allows us to determine the edges of the
defect, the centre of mass of the defect and the perimeter of the area describing the defect. These
features are shown in the figure below as additional parameters describing the defect, and are used
in the quantitative
Figure
Figure 7. assessment
Measurementof
7. Measurement of the
of the defect.
the height
height deviation
deviation distribution
distribution in
in relation to the
relation to the reference
reference plane.
plane.

The shape and area of the area outside the tolerance is determined by comparing the height of
each pixel that creates the actual surface image with the height value of the reference plane. It is a
point transformation allowing us to determine the surface area of the defect area. An example of the
defect area may be found in Figure 8. Image analysis also allows us to determine the edges of the
defect, the centre of mass of the defect and the perimeter of the area describing the defect. These
features are shown in the figure below as additional parameters describing the defect, and are used
in the quantitative assessment of the defect.

Figure 8. 8.
Figure Analysis
Analysisofofshape
shapeand
andmeasurement of the
measurement of thesurface
surfacearea
areaofofthe
the defect.
defect.

AllAll
in in
all,all,
thethealgorithm
algorithmofoflocation
location and
and assessment
assessment of ofdefect
defectparameters
parametersused usedinin
thethe
presented
presented
example includes the analysis of point-forming surfaces, determination of the reference surface and
example includes the analysis of point-forming surfaces, determination of the reference surface and
determination of areas outside the assumed tolerance of surface flatness deviation. This algorithm also
determination of areas outside the assumed tolerance of surface flatness deviation. This algorithm
allows locating delamination and defects associated with losses and material allowances observed on
also allows locating delamination and defects associated with losses and material allowances
the surface of the tile. Figure 9 shows the delamination image visible as a separation and elevation of a
observed on the surface of the tile. Figure 9 shows the delamination image visible as a separation and
part of the surface as a result of the stress stored in the ceramic tile material during the pressing and
elevation of a part of the
Figure 8. surface
Analysis as a result
of shape andof the stress stored
measurement in the area
of the surface ceramic
of thetile material during the
defect.
firing operation of the material. The area of delamination is about 50% of the entire tile area.
pressingTheand firing operation
reference of the material.
plane determined The of
on the basis area
theof delamination
tile is about
surface analysis allows50% ofindicate
us to the entire
thetile
area. All in all, the algorithm of location and assessment of defect parameters used in the presented
delamination edge and the description of the defect area in the form of delamination. The defect area
example includes the analysis of point-forming surfaces, determination of the reference surface and
isolated during the 3D image analysis is presented in Figure 10. In the three-dimensional image, apart
determination of areas outside the assumed tolerance of surface flatness deviation. This algorithm
from the delamination defect, there is also a defect of parallelism in the form of the raised right edge of
also allows locating delamination and defects associated with losses and material allowances
the tile relative to the reference plane and a small area in the lower left corner of the tile.
observed on the surface of the tile. Figure 9 shows the delamination image visible as a separation and
elevation of a part of the surface as a result of the stress stored in the ceramic tile material during the
pressing and firing operation of the material. The area of delamination is about 50% of the entire tile
area.
Materials 2020,
Materials 2019, 13,
12, 1250
x FOR PEER REVIEW 10 of
10 of 13
13

Figure 9. Example of surface flaws in the form of delamination defect on the surface of about 50% of
the tile field.

The reference plane determined on the basis of the tile surface analysis allows us to indicate the
delamination edge and the description of the defect area in the form of delamination. The defect area
isolated during the 3D image analysis is presented in Figure 10. In the three-dimensional image, apart
fromFigure
the delamination
9. Example ofdefect,
surfacethere
flawsisinalso a defect
the form of parallelism
of delamination in the
defect form
on the of the
surface of raised right
about 50% of edge
Figure
of thethe
tile 9. Example of surface flaws in the form of delamination defect on the surface of about 50% of
tilerelative
field. to the reference plane and a small area in the lower left corner of the tile.
the tile field.

The reference plane determined on the basis of the tile surface analysis allows us to indicate the
delamination edge and the description of the defect area in the form of delamination. The defect area
isolated during the 3D image analysis is presented in Figure 10. In the three-dimensional image, apart
from the delamination defect, there is also a defect of parallelism in the form of the raised right edge
of the tile relative to the reference plane and a small area in the lower left corner of the tile.

Figure 10.
Figure 10. Example
Example of
of surface
surface flaws in the
flaws in the form
form of
of delamination
delamination fault on the
fault on the surface
surface of
of about
about 50%
50% of
of
the tile.
the tile.

The measurement
The measurementofof flatness
flatness deviation
deviation associated
associated with delamination
with delamination is also performed
is also performed by scanningby
scanning all the image points in the area of the tile. Figure 11 shows three sample height profiles set
all the image points in the area of the tile. Figure 11 shows three sample height profiles set on the surface.
on the
For eachsurface.
profile,Fortheeach profile,
height the height
deviation deviationrelative
was measured was measured relative to
to the reference the reference
surface surface
at each point of
at each point of the profile. Then, for each profile, the distribution of height deviation was plotted.
the profile. Then, for each profile, the distribution of height deviation was plotted. The operation of
the Figure 10.
Theanalysis
operation of Example
of the
the of surface
entire surface
analysis offlaws
the
of inceramic
the form
entire
the of delamination
surface
tile of the
and theceramicfaulttile
on and
determination the of
surface
the of about
thedetermination
height 50% of
of the
deviation for
height
the the
entiretile.
deviation for the
defect area wasentire defect area
completed withinwas completed
200 ms. The within
area of 200
the ms. Thewas
defect area of the defect
determined bywas
the
determined by the surface area and location of the centre of mass of the defect.
surface area and location of the centre of mass of the defect.
The measurement
Based on the determined of flatness deviation
parameters, associated
the defect with delamination
is assessed and classified.isDefects
also performed by
are qualified
scanning all the image points in the area of the tile. Figure 11 shows three
on the basis of the adopted parameters that describe their acceptable dimensions as prepared by the sample height profiles set
on the surface.
quality inspection Fordepartment
each profile, ofthe
theheight deviation
production unit. was measured relative to the reference surface
at each pointthe
During of the profile.
research Then,imaging
phase, for eachofprofile,
a set ofthe50distribution
ceramic tiles of selected
height deviation was plotted.
by the manufacturer
with various surface defects was performed. A set of typical defects found in the production ofoftiles
The operation of the analysis of the entire surface of the ceramic tile and the determination the
height deviation for the entire defect area was completed within 200 ms.
in various dimensions of defects was selected. Each defect was tested in four plate settings, whichThe area of the defect was
determined
gave by the images
200 different surface of areatheand location
defects of the centre
depending on theof scan
massdirection
of the defect.
of the plate. Subsequent
image recordings were made using several sets of imaging parameters: power change, laser mode
(continuous pulse) and changes in the frequency profile registration frequency. The set of test images
has also been supplemented with tiles surfaces made correctly. The set of images prepared in this
Materials 2020, 13, 1250 11 of 13

way (about 400) was analysed several times with various sets of parameters used in the 3D image
evaluation algorithm: changes in the image filtering method, changes in the range of pixels used to
determine the reference plane, changes in the method of averaging the pixel height on scanning lines,
etc. Preparing the solution in the years 2017–2019, the sets of defects were supplemented, and the
algorithm was updated. In addition, tests were carried out in the production line working conditions
Materials 2019,the
confirming 12, xeffectiveness
FOR PEER REVIEW
of the system work. 11 of 13

Figure 11.
Figure 11. Location
Location of
of delamination
delamination defects
defects and
and determination of the
determination of the distribution of height
distribution of height deviation
deviation
from the reference plane.
from the reference plane.

Based
The on thethreat
biggest determined parameters,
to the system the defect
is the need is assessed
to shorten and classified.
the analysis Defects
time associated areincreasing
with qualified
on the
the basis of
efficiency ofthe
the adopted parameters
production line. This that describe
involves theirtoacceptable
the need reduce thedimensions as prepared
number of analysed by the
measuring
quality inspection department of the production unit.
points on the 3D surface image. In this case, microcracks with dimensions below 0.25 mm may not
During the research phase, imaging of a set of 50 ceramic tiles selected by the manufacturer with
be visible.
various surface defects was performed. A set of typical defects found in the production of tiles in
4. Discussion
various dimensions of defects was selected. Each defect was tested in four plate settings, which gave
200 different images
Studies have of the
shown thatdefects depending
it is possible on the scan
to imaging direction
typical defects of the plate.
arising Subsequent
on ceramic image
tile surfaces.
recordings were made using several sets of imaging parameters: power change, laser mode
It is also possible to describe the parameters of the defect in the form of its geometric dimensions,
(continuous
surface area pulse) and changes
and volume. in the
It is also frequency
possible profile registration
to determine the heightfrequency.
profile in anyThecross-section
set of test images
and
has also been
calculate supplemented
the deviation of flatnesswithoftiles surfacesAll
the surface. made correctly. The
measurements are set of images
carried out withprepared in this
the resolution
way (about
assumed 400) the
during wasconstruction
analysed several of thetimes with various sets
three-dimensional of parameters
vision system. These usedareinthe
theresolutions
3D image
evaluation algorithm:
determined on each axischanges
of the in the imagestation.
measuring filtering
Themethod,
points changes in the
describing range of designated
the surfaces pixels used byto
determine
the the system
3D vision reference
areplane, changes
subjected in the method
to statistical of averaging
analysis thedetermine
in order to pixel height theon scanning
next lines,
parameters
etc. Preparing
required by thethe solution
system user.inThethe article
years 2017–2019, the setswith
presents a system of defects weredX
resolution supplemented,
= dY = 0.25 mm andand
the
algorithm
dZ was updated.
= 0.35 mm. In addition,
The measurement tests were
resolution cancarried out in the
be increased production
by using line working
correspondingly conditions
larger sensor
confirming
matrices, theSony
e.g., effectiveness
IMX304 type of thewith
system work. of 4112 × 3009. Resolution measurements for the
a resolution
The biggest
presented 3D visionthreat
systemto configuration
the system isusing the need to shorten
selected matrices the analysisin time
are shown associated
the Figure with
12 below.
increasing the efficiency of the production line. This involves the need to reduce the number of
analysed measuring points on the 3D surface image. In this case, microcracks with dimensions below
0.25 mm may not be visible.

4. Discussion
Studies have shown that it is possible to imaging typical defects arising on ceramic tile surfaces.
It is also possible to describe the parameters of the defect in the form of its geometric dimensions,
surface area and volume. It is also possible to determine the height profile in any cross-section and
Materials 2019, 12, x FOR PEER REVIEW 12 of 13

required by the system user. The article presents a system with resolution dX = dY = 0.25 mm and dZ
= 0.35 mm. The measurement resolution can be increased by using correspondingly larger sensor
matrices, e.g.,
Materials 2020, 13,Sony
1250 IMX304 type with a resolution of 4112 × 3009. Resolution measurements for the
12 of 13
presented 3D vision system configuration using selected matrices are shown in the Figure 12 below.

Figure
Figure 12.
12. Measurement
Measurementresolution
resolutionfor
forselected
selected sensor
sensor matrices.
matrices.

ItIt is possible to
is possible tomake
makemeasurements
measurements at at higher
higher resolutions
resolutions withwith a narrow
a narrow field field
of viewof of
view
the of the
vision
vision system and the use of sensor matrices with higher resolution. A system of several vision
system and the use of sensor matrices with higher resolution. A system of several vision systems can
systems
be used can be usedthe
to illustrate to illustrate
surface ofthe
thesurface
ceramicoftile.
theHowever,
ceramic tile. However,conditions,
in industrial in industrial conditions,
vibrations and
vibrations and other types of noise occur on production lines that prevent imaging with increased
other types of noise occur on production lines that prevent imaging with increased resolution.
resolution.
5. Conclusions
5. Conclusions
The use of three-dimensional image analysis in the field of searching for, locating, and assessing
surface
Thedefects of ceramic tiles in the
use of three-dimensional image form of surface
analysis in thedefects
field ofsuch as cracks
searching and
for, delamination
locating, allows
and assessing
effectivedefects
surface assessment of thetiles
of ceramic quality of the
in the form surface of thedefects
of surface tile. Insuch
addition, the use
as cracks andofdelamination
a three-dimensional
allows
image for fast and industrial surface evaluation in the scope of its geometry
effective assessment of the quality of the surface of the tile. In addition, the use of a three-dimensional allows for integration
and extension
image for fast and of control
industrial tasks and evaluation
surface their implementation
in the scopeatofone control station.
its geometry allows for Theintegration
presented
method
and of measuring
extension of control thetasks
flatness
and of ceramic
their tiles using aatthree-dimensional
implementation one control station. image containingmethod
The presented 800,000
measurement points allows us to determine a map of the distribution
of measuring the flatness of ceramic tiles using a three-dimensional image containing 800,000 of flatness defects on the tile
surface and topoints
measurement check each
allows product within about
us to determine 200 ms.
a map of theHowever, it does
distribution of not include
flatness image
defects onbuilding
the tile
surface and to check each product within about 200 ms. However, it does not include image extension
time and communication time with executive devices (machine controllers). It is a significant building
of inspection
time capabilities in time
and communication relation
with to the sensor systems
executive devices installed
(machine oncontrollers).
the line enabling It ismeasurements
a significant
at several points on the surface of the tile.
extension of inspection capabilities in relation to the sensor systems installed on the line enabling
The 3D vision
measurements systempoints
at several is designed
on theto perform
surface of athe
specific
tile. inspection task. One of the most important
parameters
The 3D vision system is designed to perform resolution.
is to achieve the required measurement The presented
a specific inspection task.system
One ofenables
the mostthe
registrationparameters
important of a defect image with dimensions
is to achieve the required of 0.25 × 0.25 mm, which
measurement gives The
resolution. 16 measurement
presented systempoints
for every square millimetre of the tile surface. This is the limit
enables the registration of a defect image with dimensions of 0.25 × 0.25 mm, which gives 16 dimension of the defect that can be
observed in apoints
measurement three-dimensional
for every square image. If needof
millimetre arises to increase
the tile the measurement
surface. This resolution,
is the limit dimension the
of the
design requirements should be changed, and other components of
defect that can be observed in a three-dimensional image. If need arises to increase the measurement the system should be selected.
The analysis
resolution, theofdesign
defect requirements
areas on the 3D imagebeallows
should changed,a repeatable
and other description
components of the
of thedefect parameters
system should
and their classification, enabling quick product checking and sorting.
be selected. The analysis of defect areas on the 3D image allows a repeatable description of the defect
The method
parameters and their of evaluation
classification, of ceramic
enablingsurface defects presented
quick product checking and in this paper, based on the 3D
sorting.
image analysis, enables the implementation of a much larger
The method of evaluation of ceramic surface defects presented in this paper, based on number of measurements andthetests,
3D
especially in the field of spatial geometric measurements. It significantly
image analysis, enables the implementation of a much larger number of measurements and tests, extends the possibilities of
detailed surface evaluation in relation to analysis using vision systems analysing
especially in the field of spatial geometric measurements. It significantly extends the possibilities of colours or shades of
grey of the
detailed tile surfaces.
surface evaluation In addition,
in relation it significantly
to analysis using expands the systems
vision detailed analysing
microdamage analysis
colours of the
or shades
surface using 16 measuring points per 1 square millimetre of a ceramic
of grey of the tile surfaces. In addition, it significantly expands the detailed microdamage analysis oftile surface.
the surface using 16 measuring points per 1 square millimetre of a ceramic tile surface.
Funding: This research received no external funding.
Acknowledgments:
Acknowledgments: The The project
project was
was implemented
implemented asas part
partof
ofresearch
researchand
anddevelopment
developmentworkworkcarried
carried out
out asas part
part of
of industrial
industrial cooperation
cooperation withwith Department
Department of Process
of Process ControlControl
workingworking within
within the the structures
structures of Faculty of Faculty of
of Mechanical
Mechanical
EngineeringEngineering and
and Robotics in Robotics in AGHof
AGH University University of Science
Science and and Technology,
Technology, Krakow, Poland.
Krakow, Poland.
Conflicts of Interest: The authors declare no conflict of interest.
Materials 2020, 13, 1250 13 of 13

References
1. Ozkan, F.; Ulutas, B. Use of an eye-tracker to assess workers in ceramic tile surface defect detection.
In Proceedings of the 3rd International Conference on Control, Decision and Information Technologies
(CoDIT), St. Paul’s Bay, Malta, 6–8 April 2016; pp. 88–91.
2. Hosseinzadeh, H.S.; Ahmad, A.; Farshad, B. Automatic detection and classification of the ceramic tiles
surface defects. Pattern Recognit. 2017, 66, 174–189.
3. Hocenski, Z.; Matic, T.; Vidovic, I. Technology Transfer of Computer Vision Defect Detection to Ceramic Tiles
Industry. In Proceedings of the 1st International Conference on Smart Systems and Technologies, Osijek,
Croatia, 12–14 October 2016; pp. 301–305.
4. Samarawickrama, Y.C.; Wickramasinghe, C.D. Matlab based Automated Surface Defect Detection System
for Ceremic Tiles using Image Processing. In Proceedings of the 6th National Conference on Technology
& Management (NCTM)—Excel in Research and Build the Nation, Malabe, Sri Lanka, 27 January 2017;
pp. 34–39.
5. Vaideliene, G.; Valantinas, J.H. Haar Wavelet-Based Approach to Locating Defects in Texture Images.
In Proceedings of the 21st International Conference on Information and Software Technologies (ICIST),
Druskininkai, Lithuania, 15–16 October 2015; Sponsor(s) Kaunas University Technol Information and
Software Technologies, ICIST 2015; Communications in Computer and Information Science Series 2015.
Volume 538, pp. 537–547.
6. De Araújo, M.R.E.; Martínez, J.; Ordóñez, C.; Vilán, J. Identification of granite varieties from colour spectrum
data. Sensors 2010, 10, 8572–8584. [CrossRef] [PubMed]
7. López, M.; Martínez, J.; Matías, J.; Vilán, J.; Taboada, J. Application of a hybrid 3D-2D laser scanning system
to the characterization of slate slabs. Sensors 2010, 10, 5949–5961. [CrossRef]
8. Iglesias, C.; Martínez, J.; Taboada, J. Automated vision system for quality inspection of slate slabs. Comput.
Ind. 2018, 99, 119–129. [CrossRef]
9. Martínez, J.; Lopez, M.; Matías, J.; Taboada, J. Classifying slate quality using automated learning techniques.
Math. Comput. Model. 2013, 57, 1716–1721. [CrossRef]
10. Olszyna, G.; Sioma, A.; Tytko, A. Assessment of the condition of hoisting ropes by measuring their geometric
parameters in a three-dimensional image of their surface. Arch. Min. Sci. 2013, 58, 643–654.
11. Sioma, A.; Socha, J. Automation of annual tree increment measurement using vision system. Drewno
2016, 59, 19–30.
12. Sioma, A. Assessment of wood surface defects based on 3D image analysis. Wood Res. 2015, 60, 339–350.
13. Sioma, A. The estimation of resolution in 3D range image system. In Proceedings of the 2013 14th International
Carpathian Control Conference, Rytro, Poland, 26–29 May 2013; pp. 346–349.
14. Lenty, B. Machine vision system for quality control of molded plastic packaging. In Proceedings of the 2019
Photonics Applications in Astronomy, Communications, Industry, and High-Energy Physics Experiments,
Wilga, Poland, 26 May–2 June 2019; SPIE: Bellingham, WA, USA, 2019.
15. Lenty, B.; Kwiek, P.; Sioma, A. Quality control automation of electric cables using machine vision.
In Proceedings of the 2018 Photonics Applications in Astronomy, Communications, Industry, and High-Energy
Physics Experiments, Wilga, Poland, 3–10 June 2018; SPIE: Bellingham, WA, USA, 2018.
16. Sioma, A.; Kowal, J.; Karwat, B. Methods of 3D imaging used in quality inspection systems in manufacturing.
J. Mach. Constr. Maint. 2017, 1, 9–17.
17. Sioma, A. Filtering algorithms for 3D range image analysis. Solid State Phenom. 2013, 199, 235–240. [CrossRef]

© 2020 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).