high-resolution x-ray imaging and analysis of coatings on and in wood. - high resolution document camera

Wood coatings are widely used in beauty and protection.
In order to build a knowledge base for product optimization, it is critical to evaluate the coating performance during service life.
In order to analyze the behavior of akating, a large number of technologies are available, and microimaging is an important tool.
In addition to the standard microscope technology, highresolution X-
X-ray tomography is a way to provide non-destructive homogenization of the substrate for the coating and its application.
Combined with the analysis of 3D volume data, the surface roughness, structure and thickness, penetration depth and related mechanical anchoring of the coating can be studied in relation to the underlying substrate.
To clearly illustrate the possibilities and limitations of this technique, opaque water-based coatings and opaque water-based coatings applied to two different wood types were scanned and analyzed. Clearly, three-dimensional X-
High-resolution ray imaging can produce valuable information only through visualization.
In addition, through appropriate analysis, X-
CT and automatic image processing.
Keywords 3D, X-
Despite the use of synthetic alternatives, the importance of wood as an industrial material cannot be underestimated.
As a renewable material with unique properties, Wood has an undeniable advantage, but the disadvantage of wood is that it has inherent variability due to its biological properties and is susceptible to microorganisms(1)
In order to protect the wood from physical and biological weathering, the application of protective coating is a good choice
In addition to wood improvement and wood preservation, wood protection techniques have been established.
The performance of wood coatings in field or simulation services is measured in a number of ways, in which micro-imaging and accompanying analysis of visual information is a way to evaluate their weathering behavior.
Conventional optical microscopy, fluorescence microscopy and other technologies (2)
And high
As the end of the atomic force microscope ,(3), (4)
Scanning electron microscopy ,(5)and con-
Focal laser microscope(6)
Only the last technique can conceive the coating in a 3D range without damage, however, although confocalmicroscopy has a higher resolution, its detection capability is still limited.
Except for the above-
Technology mentioned, X-
Ray tomography is a high
Fastand non-destructive three-end tool
Size analysis.
This technology has been used in plant biology (7)
Soil Science. (8)
Previous research using X
A ray tomography was performed on the thermal insulation coating ,(9)paper,(10), (11)
Cultural Heritage. (12)High-resolution X-
Ray tomography is particularly interesting to study the structure of the wood coating and the average thickness and penetration depth of the coating.
In addition, the digital coating structure can also analyze its surface and internal structure.
Show the power of X
X-ray tomography and its limitations, visualization and analysis of two coated wood samples were carried out.
Description of wood using experimental methods and material samples of the following tree species: Scottish pine (Pinussilvestris)and padouk (
Rosewood soyauxii)
Representing the local cork and tropical hard wood.
Pine sapwood is a lightweight material commonly used in European standards as a test substrate, and aspadouk is a durable and lush tropical species from Africa for the production of window carpentry.
The wood is cut straight.
Saw the granular material and two small plates with a growth ring angle of 45 [degrees]
With the surface.
The board is brushed.
It is coated with opaque solvent and opaque water-based paint.
Details of coating application and coating composition are not given as the purpose of this article is to illustrate X-
Ray imaging and analysis, not internal analysis
Deep study of wood coating itself.
By sawing a circuit board into a parallel six-pole board, an image acquisition sample for scanning was preparedSpecimen of shape
Measuring the top of a wood sample of about 2x2x2mm with X-scan
Ray equipment built in center X
Ray tomography at GhentUniversity (UGCT;
This is a state. of-the-artscanner, (13)
Highly flexible
House has developed software for scanner control, sample reconstruction, analysis and visualization. TheX-
Ray source, Fein-focus nano-
Focus tube, can reach the focus of 1 [size]micro]m.
All samples are scanned at an average voltage of 70 kV with a target current of 30 [micro]
A, the exposure time is MS On 1800.
Rotation step of 0. 36[degrees]was used.
Octopus is a server/client fault scan reconstruction package for parallel and cone beam geometry with a reconstruction time of 20 minutes (www. xraylab. com). (14)
Settings with description-
In this paper, micron-level resolution can be achieved, resulting in a scan with a pixel size of about 2x2x2 [micro]mwith 2 (16)
Gray level.
Image analysis loading image in MATLAB [R]
Used for pre-processing and analysis.
The purpose is to quantify the surface roughness, thickness and structure of the coating, contact surface and penetration depth.
Each step of the analysis is described below.
Although the image quality of the reconstructed slice has been improved very well, bilateral filtering as an edge
Keep the smoothing technique to prevent the average edge and still average in the smoothing area (15)
An improvement for image segmentation.
Segmentation is almost always the most critical part of image analysis. (16)
Due to the difference of X-
Ray density of coating, wood and noise, bilateral filtering scanning can be done by fitting by three Gaussian-
The shape of the curve to the histogram of the Cross
Segments of pine trees.
Figure 1 shows the histogram of the resized slices, in line with the normal distribution of pine trees.
The intersection between the first and second Gaussian distributions will separate the noise from the wood.
The intersection of the second and third Gaussian curves is used as a threshold for separating the coating from wood.
Obviously, the difference between the wood and the coated body is more obvious than the boundary between the noise and the wood. [
Figure 1 slightly]
Shape closure and size of cleaning combination-
Based on the opening of the split volume, isolated voxels or voxels were removed and incorrectly classified as coatings due to imperfect segmentation.
When scanning, the sample is placed vertically and the coating is parallel to the horizontal plane. However, in order to correctly calculate the layer thickness and penetration, a small deviation from the lateral position must be Mathematically corrected by the rotation of the volume.
Therefore, the surface of the coating is extracted from said volume and the fitting plane is used by principal component analysis.
Therefore, the normal of the plane gives the rotation angle required for the data space transformation.
To further analyze the extracted and rotating surfaces, the height averaged zero, and several roughness parameters were calculated using image j using the SurfCharJ toolbox of ch et al. (17)
: Arithmetic Mean Deviation ([R. sub. a])
Root of mean variance ([R. sub. q])
, Evaluate the peak of the configuration file ([R. sub. ku])
, To evaluate the degree of deviation of the contour ([R. sub. sk])
Lowest valley[R. sub. v]), highestpeak ([R. sub. p])
The total height of the outline ([R. sub. t]). Polarangles (orientation)
And azimuth (direction)
The faces are also calculated according to the face normal of the triangular surface.
In order to rule out that the coating holes are incorrectly included in the penetration depth calculation, the morphological filling of the coating holes is mandatory.
Two methods are used to calculate the layered thickness and penetration depth from the vertical down surface.
It is stated in figs.
Figure 2b and original image in 2c2a. [
Figure 2:
In theory, penetration does not take into account the simple filling of cells that are cut on the surface.
However, the range of coating penetration is controlled by the ability to flow into the opening and by the transport of the interconnect pit. (18)
Figure 2 according to the theoretical definition, the division of layer thickness is given according to Method 1.
The remaining coating under the coating is considered to penetrate into the substrate.
Method 2, as shown in the figure.
2c, similar to the method described by Van denBulcke et al. (19)
Based on two-
The dimension view of the penetration layer thickness includes all coating materials above the surface roughness and all materials below the penetration layer.
The calculation of the hole rate of the segmented coating volume is based on the hole volume distribution and the maximum internal cutting ball distribution.
Interface of the interface coating, I. e.
, The contact surface between the coating and the wood, by calculating the area of the triangular interface surface as an approximation of the real surface area, is quantified as a measure of adhesion.
Logically, the fineness of the mesh determines the accuracy of the surface approximation.
The reconstruction and analysis images of all samples are implemented with VGStudio MAX [R], MATLAB[R]
, ImageJ and Drishti. (20)
Results and Discussion Figure 3 illustrates several Twoand three-
Represents the size image of the volume rendering and analyzes the data of opaque water and opaque waterborne coating. [
Figure 3 slightly]
Figure 3a and 3b are original scans, fake colors-
Manual segmentation based on histogram.
By virtually removing a portion of the wood base, the imprint on the wood surface at the base of the coating is clearly visible.
As expected, the casting of the coating to the Wood revealed that the rough printing of forpadouk was significantly less than that of pine trees.
Insert in Fig.
3a and 3b are a detailed crossover
Slices of coated wood samples, similar to images obtained with a co-focused microscope (6), (19)
It can be used for manual calibration of formation thickness and penetration depth.
Obviously, the high density of padouk results in a smaller contrast between the coating and the wood, interfering with the segmentation.
Figure 3c and 3d, with artificial products ,(21)
The extraction surface of the coating is described with an average roughness of zero.
Of course, the level of detail is determined by the size of the body of the scan, which is sufficient for wood carving.
Table 1 lists several roughness parameters calculated for the two coatings under test.
The observation that the surface roughness of coated pine trees is higher than that of paddock is self-
Obviously, the padouksurface is usually smoother.
However, it should be emphasized that these ceremonies take place in about 2 [Square]mm. sup. 2]
The coated padouk does not represent the entire plate due to the heterogeneity of the wood.
Areas including ships are not included in the scan but may have a significant impact on the coating structure.
In contrast, the surface roughness of the pine tree is more or less representative for the entire board surface, at least for the early wood areas that became apparent in the pine scan.
Usually, by the early-wood-
Due to incomplete leveling, late Wood differences occur in coatings with medium thickness.
An interesting view is given by the direction and direction of the surface.
Figure s3e and 3f show directions, fig.
The Forpine and padouk directions are 3g and 3 h.
For both coatings, due to the high slope, there is only a high angle at certain specific points, and the orientation is the inherent low of the flat surface.
The orientation image more or less reflects the terrain seen in figs. 3c and 3d.
This information can be related to gloss values and can be implemented in light models, such as the reflective Beckman theory for gloss prediction.
Especially when changes are monitored during weathering, surface representation is very important. (4)
In addition, the evolution of small cracks on the surface of the coating indicated by black and white arrows (Figs. 3c and 3e)
It can be monitored throughout the weathering process.
In this regard, the non-destructive nature of technology is an obvious advantage.
The determination of formation thickness and penetration depth is more complicated.
In order to quantify this effect, it is necessary to be able to properly segment the coating and wood at the interface, and this method is feasible only if the gray level is significantly different.
According to Method 1, the average layer thickness and penetration depth are 58, respectively. 2 and 3. 8 [micro]
Pine and 43 metres. 8 and 2. 5 [micro]m for padouk.
By definition, the value of permeability is quite low.
However, according to Method 2 given by Van den Bulcke et al. , (19)
The results are expected to be very different, but the real Three
The latter definition does not make full use of the potential of dimensional information.
In this case, the layer thickness and penetration depth are 36, respectively. 7 and 21. 4[micro]
Pine and 37 metres. 5 and 6. 1 [micro]m for padouk.
These results are more or less consistent with the results found by Van den Bulckeet al. (19)
Similar coating. The two-
According to the first method described in figure 1, the dimensions of the layer thickness and penetration are illustrated.
Only 4a and 4b of pine trees. [
Figure 4 slightly]These two-
Size illustrations are reasonable for pine trees, but forpadouk misclassification results in a wrong removal of the coated voxelor, or an increase in height
Density crystals present in the wood substrate into the coated substrate.
Although the mean smoothed such fault segmentation and is a good estimate of the thickness of the layer, for padouk, it is not accurate enough to accurately determine the penetration of each position.
Especially the automation of penetration computing, there is still a lot of work to be done.
To eliminate incorrect segmentation, a solid solution when detailed study penetration is the target is differential imaging: subtraction of images before and after coating application.