Presentation is loading. Please wait.

Presentation is loading. Please wait.

S. MEZGHANI, E. PERRIN, V. VRABIE

Similar presentations


Presentation on theme: "S. MEZGHANI, E. PERRIN, V. VRABIE"— Presentation transcript:

1 S. MEZGHANI, E. PERRIN, V. VRABIE
Multiscale characterization of undercoat alteration using active infrared thermography S. MEZGHANI, E. PERRIN, V. VRABIE CReSTIC-Châlons, EA 3804, IUT de Reims-Châlons-Charleville, Chaussée du Port, B.P. 541, Châlons- en-Champagne, France M. Z. Ahmad, A. A. Khan School of Electrical Engineering and Computer Science, National University of Sciences and Technology, Islamabad, Pakistan J.L BODNAR GRESPI/ECATHERM, Université de Reims Champagne-Ardenne, BP 1039, Reims, France B. CAUWE CRITT-MDTS, 3 Boulevard Jean Delautre - ZHT du Moulin Leblanc, Charleville-Mézières, France Good morning lady and gentleman. I m sihem mezghani, thanks you to introduce me. This presentation outline our work on multiscale analysis of thermography imaging dynamic for sol-gel coating discrimination undertaken by collaboration between GRESPI and industrial research laboratory CRITT. The objective of our project is the characterization of surface coating and discriminate the sol-gel sub-layer presence among a mass of sample using thermography techniques.

2 Plan Context IR thermography technique Thermal enhancement approach
Wavelets pre-treatment of raw thermal data Selection of pertinent sub-space Choice of the Wavelets regarding excitation/discrimination Industrial application Detection of undercoat Defect detection Conclusions et perspectives So, I will start this presentation by a brief presentation of the context and objective of this study, then I will expose the IR thermography experiment carried in this work and followed by a discussion of results of cross correlation analysis and wavelet pre-processing to sol-gel coating discrimination. We interest particularly to the effect wavelet function choice regarding discrimination performance. I finish this presentation by some conclusions and perspectives.

3 Context Characterization of undercoat alteration:
Detection of undercoat (sol gel coating); Coating and undercoat characterization Sol-gel method advantage Excellent adhesion between the metallic substrate and the top coat. Sol-gel Coated samples Corrosion protection. Paint Applications Sol-gel Simple, economic and effective coating method Automotive Aerospace industries Substrate To achieve a surface functionality, surface coating operation is adopted in several industrial applications. In particular, paint spray and sol gel prcess well known as simple to implant, take place under mild condition. This double layer is applied in several application as: Aeronautics which the sol gel solutions is used for repair of impacts suffered by aircraft. In the field of automotive painting the sol gel process relate more transparent part :glass, optic, the advantage of this technology over conventional method is the reduction of time and production of time and production cost. Presence of undercoat

4 Context Characterization of undercoat alteration:
Defect detection for different materials Material inspection Defected sample Applications Paint Substrate Artwork inspection Mechanical industries Aerospace industries Civil structures Detection of Subsurface Defects Thickness variation of undercoat

5 Different techniques for characterization of coated surfaces
Context Different techniques for characterization of coated surfaces Cross section There are a need of non destructive method for large inspected area Destructive Infrared thermography Ultrasonic test Non-destructive for coating quality inspection, two kinds of techniques can mainly be differentiated : non destructive and destructive testing. NDT is the process of inspection and evaluating material component without the service ability of the sample. In contrast to NDT, other tests are destructive in nature and are therefore done on a limited number of samples. This study is devoted to validate a new NDT using thermograph experiment for evaluating the paint thickness variation. the advantage of Pt compared to other NDT technique is the cost associated to IR equipement, fast inspection which only fe’w second are required to perform the acquisition and analysis,

6 IR Thermography technique
Configuration setup Data acquisition Sample to be inspected: Single-layered coated sample; Bi-layered coated sample; Temperature (K) IR camera PC/Analysis: Recording; Synchronisation Y(pixels) X(pixels) Recording thermal image The setup consist of an IR camera, a manual flash lamp and a computer. As we can see in this video recorded sequence, the temperature profile of solgel coated sample show more high deviation from ambient temperature at different point than observed in new coated sample. All this temporal image sequence is stored into a single cube structure of data Photographic heat source: Halogen; Flash; Sequence of thermal image 3D thermal cube data

7 Thermal enhancement approach:
Pre-processing analysis of raw thermal data Thermal image At Discrete Wavelets transform As the same way where the Newton prism decomposes the relief of surface in different wavelength, wavelet transform are multi-resolution image decomposition tool that provide image feature at different scale sub-band by applying high and low pass filter, it provided by the determination of a function Ø at different resolution. The approximation and detail component is separated at each decomposition level. Regarding to 2DWT we can consider detail denoted CDk detail is the difference between two successive approximation levels.

8 Thermal enhancement approach:
D1 Selection of pertinent sub-space Surface roughness Global 3D Raw thermal image Di Di+1 Useful information Y(pixels) Dk Recording thermal image X(pixels) The wavelet function is impacted by: Duration of pulsed excitation; Application; Quality of the discrimination; Dk+1 Surface shape This multiscale decomposition is then applied for all recorded data represented here by a cube. The decomposition is occurred in this study in seven levels. Dn

9 Choice of the Wavelets regarding discrimination
2 4 8 2 4 8 2 4 8 D4 D5 D6 D7 2 4 8 2 4 8 2 4 8 2 4 8 X-direction is sorted by different wavelet orders : 2, 4 and 8 Different families Lets now see the impact of wavelet function choice on the discriminate rate. Four wavelet functions were tested. We remark here that between the selected discriminative components D4,D5 and D6, D5 is less sensitive to the choice of wavelet function. Daubechies The wavelet function does not impact the quality of the discrimination at level D5. Symmlets Biorthogonal Reverse Biorthogonal

10 Choice of the Wavelets regarding excitation
Times Te Wavelet’s mother function Pulse duration Te=0,5s or Te=10s Te=1s/2s Te=5s Coiflet Wavelets:1 Symlets Wavelets:4 ReverseBior Wavelets: 6.8 Lets now see the impact of wavelet function choice on the discriminate rate. Four wavelet functions were tested. We remark here that between the selected discriminative components D4,D5 and D6, D5 is less sensitive to the choice of wavelet function. to select the basis which provides best separation of the desired faults from the background without compromising on the weaker faults. M. Z. Ahmad, A. A. Khan, S. Mezghani, E. Perrin, K. Mohoubi, J. L. Bodnar, V.Vrabie, Wavelet subspace decomposition of thermal infrared images for defect detection in artworks, arvix 2016.

11 Application : Methodology
Selection of inspected sample Thermal excitation Data acquisition of 3D Thermal data Best raw image Application of DWT to thermal Characterization of undercoat alteration Comparison of images Selection of pertinent subspace Data reconstruction from the selected subspace Temporal average of thermal image This loops is applied to sol-gel coated sample and reference sample

12 Application : detection of undercoat
Uncoated Sol-gel coated Composition of sample: Commercial white paint Sol-gel sublayer deposited by electrostatic spray deposition Steel substrate (a) (b) Paint layer Paint layer Sol-gel layer 12µm Steel substrat In this study, ten rectangular steel samples were considered. Only five samples were coated by thin film of sol gel, and then all samples were next painted with a commercial white paint. Look at cross section micrographs of this sample obtained by optical microscopy. The sol gel layer has 12 µm thick and paint layer has almost 50µm thick. Steel substrat Cross-section micrographs by optical microscopy (Olympus BX60). Enhancement of thermal imaging to detect the hidden Sol-gel undercoat

13 Application : defect detection
Sample In front side Directly exposed to excitation No visible substrate thickness variation and substrate alterations Sample composition : Commercial black paint White plaster substrate In rear side Containing a simulated defects Underocat invisible information In this study, ten rectangular steel samples were considered. Only five samples were coated by thin film of sol gel, and then all samples were next painted with a commercial white paint. Look at cross section micrographs of this sample obtained by optical microscopy. The sol gel layer has 12 µm thick and paint layer has almost 50µm thick. Enhancement of thermal imaging to detect the invisible information

14 Application: results Non uniform heating Surface roughness Defect
Uncoated sample Sol-gel coated sample Defected uncoated sample Non uniform heating Surface roughness Defect Filtered image using 3D extension of 2D multiscale wavelets transform In this study, ten rectangular steel samples were considered. Only five samples were coated by thin film of sol gel, and then all samples were next painted with a commercial white paint. Look at cross section micrographs of this sample obtained by optical microscopy. The sol gel layer has 12 µm thick and paint layer has almost 50µm thick. Overcoming of the background related to pictorial layer and removes the undesired high frequency noise represented in experimental noise .

15 Conclusions The multiscale wavelet decomposition improve the analysis of recorded thermal images The meaningful appropriate subspaces provide pertinent information about the undercoat alteration The characterization of undercoat alteration is enhanced by overcoming of the background related to pictorial layer and removes the undesired high frequency noise present in experiments Adaptation of wavelet choice regarding the duration of excitation and choice of discrimination level To conclude, This result show that the wavelet decomposition allows an enhancing detection of the undercoat alteration in material structure by selecting details levels better corresponding to the thermal signature of undercoat thickness variation. it seems possible to characterize the thermal change of undercoat alteration using active infrared thermography. To assure the performance for various industrial applications, we are actually testing the effect of wavelet choice and excitation type in the detection of undercoat alteration as was proposed recently by Ahmad et al.

16 Thanks for your attention Questions ?


Download ppt "S. MEZGHANI, E. PERRIN, V. VRABIE"

Similar presentations


Ads by Google