Papers by Jean-baptiste Thomas
The visual systems found in nature rely on capturing light under different modalities, in terms o... more The visual systems found in nature rely on capturing light under different modalities, in terms of spectral sensitivities and polarization sensitivities. Numerous imaging techniques are inspired by this variety, among which, the most famous is color imaging inspired by the trichromacy theory of the human visual system. We investigate the spectral and polarimetric properties of biological imaging systems that will lead to the best performance on scene imaging through haze, i.e., dehazing. We design a benchmark experiment based on modalities inspired by several visual systems, and adapt state-of-the-art image reconstruction algorithms to those modalities. We show the difference in performance of each studied systems and discuss it in front of our methodology and the statistical relevance of our data.
Figure : We encounter materials permitting some degree of subsurface light transport, described a... more Figure : We encounter materials permitting some degree of subsurface light transport, described as transparent or translucent.
Logistic splicing correction for VNIR–SWIR reflectance imaging spectroscopy
Optics Letters, Jan 5, 2023
In the field of spectroscopy, a splicing correction is a process by which two spectra captured wi... more In the field of spectroscopy, a splicing correction is a process by which two spectra captured with different sensors in adjacent or overlapping electromagnetic spectrum ranges are smoothly connected. In our study, we extend this concept to the case of reflectance imaging spectroscopy in the visible–near-infrared (VNIR) and short-wave infrared (SWIR), accounting for additional sources of noise that arise at the pixel level. The proposed approach exploits the adaptive fitting of a logistic function to compute correcting coefficients that harmonize the two spectral sets. This short Letter addresses usage conditions and compares results against the existing state of the art.
Investigating the Kokhanovsky snow reflectance model in closerange spectral imaging
Final program and proceedings, Nov 1, 2021
The internal structure of the snow and its reflectance function play a major contribution in its ... more The internal structure of the snow and its reflectance function play a major contribution in its appearance. We investigate the snow reflectance model introduced by Kokhanovsky and Zege in a close-range imaging scale. By monitoring the evolution of melting snow through time using hyperspectral cameras in a laboratory, we estimate snow grain sizes from 0.24 to 8.49 mm depending on the grain shape assumption chosen. Using our experimental results, we observe differences in the reconstructed reflectance spectra with the model regarding the spectra's shape or magnitude. Those variations may be due to our data or to the grain shape assumption of the model. We introduce an effective parameter describing both the snow grain size and the snow grain shape, to give us the opportunity to select the adapted assumption. The computational technique is ready, but more ground truths are required to validate the model.
Impact of Shape on Apparent Translucency Differences
Final program and proceedings, Oct 21, 2019
Translucency is one of the major appearance attributes. Apparent translucency is impacted by vari... more Translucency is one of the major appearance attributes. Apparent translucency is impacted by various factors including object shape and geometry. Despite general proposals that object shape and geometry have a significant effect on apparent translucency, no quantification has been made so far. Quantifying and modeling the impact of geometry, as well as comprehensive understanding of the translucency perception process, are a point of not only academic, but also industrial interest with 3D printing as an example among many. We hypothesize that a presence of thin areas in the object facilitates material translucency estimation and changes in material properties have larger impact on apparent translucency of the objects with thin areas. Computergenerated images of objects with various geometry and thickness have been used for a psychophysical experiment in order to quantify apparent translucency difference between objects while varying material absorption and scattering properties. Finally, absorption and scattering difference thresholds where the human visual system starts perceiving translucency difference need to be identified and its consistency needs to be analyzed across different shapes and geometries.
Measuring the Relative Image Contrast of Projection Displays
Final program and proceedings, Oct 18, 2015
Projection displays, compared to other modern display technologies, have many unique advantages. ... more Projection displays, compared to other modern display technologies, have many unique advantages. However, the image quality assessment of projection displays has not been well studied so far. In this paper, we propose an objective approach to measure the relative contrast of projection displays based on the pictures taken with a calibrated digital camera in a dark room where the projector is the only light source. A set of carefully selected natural images is modified to generate multiple levels of image contrast. In order to enhance the validity, reliability, and robustness of our research, we performed the experiments in similar viewing conditions at two separate geographical locations with different projection displays. In each location, we had a group of observers to give perceptual ratings. Further, we adopted state-of-art contrast measures to evaluate the relative contrast of the acquired images. The experimental results suggest that the Michelson contrast measure performs the worst, as expected, while other global contrast measures perform relatively better, but they have less correlation with the perceptual ratings than local contrast measures. The local contrast measures perform better than global contrast measures for all test images, but all contrast measures failed on the test images with low luminance or dominant colors and without texture areas. In addition, the high correlations between the experimental results for the two projections displays indicate that our proposed assessment approach is valid, reliable, and consistent.
HAL (Le Centre pour la Communication Scientifique Directe), 2021
Pigment mapping allows the classification and estimation of the abundances of pigments in paintin... more Pigment mapping allows the classification and estimation of the abundances of pigments in paintings. The information learned becomes extremely important for conservators, who are then able to decide the best strategies in the conservation of the artefacts. When the goal is to restore a painting, it is also important to know what the effects of the newly introduced materials are. To fulfil this purpose, a proper mixing model must be defined. We propose a framework to perform pigment mapping on the hyperspectral image of an experimental painting realised for the occasion, with the goal of rendering a colour image using the concentrations retrieved from the mapping. Contrarily to spectral unmixing tasks, where subtractive models prevailed, hybrid models have the advantage of outputting more accurate colours in this workflow.
Journal of perceptual imaging, Mar 1, 2022
In this work we study the perception of suprathreshold translucency differences to expand the kno... more In this work we study the perception of suprathreshold translucency differences to expand the knowledge about material appearance perception in imaging and computer graphics, and 3D printing applications. Translucency is one of the most considerable appearance attributes that significantly affects the look of objects and materials. However, the knowledge about translucency perception remains limited. Even less is known about the perception of translucency differences between materials. We hypothesize that humans are more sensitive to small changes in absorption and scattering coefficients when optically thin materials are examined and when objects have geometrically thin parts. To test these hypotheses, we generated images of objects with different shapes and subsurface scattering properties and conducted psychophysical experiments with these visual stimuli. The analysis of the experimental data supports these hypotheses and based on post experiment comments made by the observers, we argue that the results could be a demonstration of a fundamental difference between translucency perception mechanisms in see-through and non-see-through objects and materials.
Image quality and perception: introduction
Journal of the Optical Society of America, May 27, 2022
This feature issue focuses on image quality and perception, including image and video quality, su... more This feature issue focuses on image quality and perception, including image and video quality, subjective and objective quality, and enhancement. The feature issue contains papers on several important topics, such as contrast discrimination, analysis of color imaging in cameras, image quality assessment, and more. The papers represent different important aspects in image quality and perception, contributing to the advancement of the field.
Behavioral Investigation of Visual Appearance Assessment
Final program and proceedings, Nov 12, 2018
Illuminant estimation from uncalibrated multispectral images
We investigate the physical validity of typical computational color constancy models for illumina... more We investigate the physical validity of typical computational color constancy models for illuminant estimation of uncalibrated multispectral images. We demonstrate empirically that the assumptions may be reasonable and that we retrieve reasonably well the illumination for some images. On these images, we also have access to a good estimate of the spectral properties of the illumination while increasing the number of bands. However, some other images do not provide a very good illuminant estimation. We also show that the result depends mostly on the scene, rather than on the hypothesis made or on the number of spectral bands. Besides, the influence of the algorithm and its hypothesis is more critical for more bands than for the 3 - D color case.
Final program and proceedings, Nov 1, 2021
Translucency optically results from subsurface light transport and plays a considerable role in h... more Translucency optically results from subsurface light transport and plays a considerable role in how objects and materials appear. Absorption and scattering coefficients parametrize the distance a photon travels inside the medium before it gets absorbed or scattered, respectively. Stimuli produced by a material for a distinct viewing condition are perceptually non-uniform w.r.t. these coefficients. In this work, we use multi-grid optimization to embed a non-perceptual absorption-scattering space into a perceptually more uniform space for translucency and lightness. In this process, we rely on A (alpha) as a perceptual translucency metric. Small Euclidean distances in the new space are roughly proportional to lightness and apparent translucency differences measured with A. This makes picking A more practical and predictable, and is a first step toward a perceptual translucency space.
Zenodo (CERN European Organization for Nuclear Research), Nov 1, 2022
In attempting to document the degradation processes occurring on cultural heritage objects, imagi... more In attempting to document the degradation processes occurring on cultural heritage objects, imaging-based analytical techniques present many advantages, as they provide spatial and spectral information and allow the simultaneous investigation of the chemical and morphological characteristics of a sample. This study presents a protocol based on chemical imaging -Fourier transform infrared microspectroscopy (μ-FTIR) and hyperspectral imaging (HSI) -aimed at monitoring the formation of metal soaps on model metal coupons. Oil-painted metal supports are in fact not immune to degradation due to metal soaps formation, a phenomenon that affects all oil-painted surfaces from the initial curing of the paint film. Copper and zinc sheets were coated with cold-pressed linseed oil and artificially aged for one month in order to instigate the formation of metal soaps. Their reaction was then monitored by means of µ-FTIR. The chemical maps showed an increasing trend over time, elucidating some aspects and differences in the mechanism of formation of the organic salts for the two metal substrates. Additionally, the samples were analysed using two hyperspectral cameras, operating in the visible-near infrared and short-wave infrared spectral range. The appropriateness of the two cameras in the investigation of metal soaps, and the effect of the thickness of the coating on the data obtained, is discussed here.
Hyperspectral Vnir - Swir Image Registration: Do Not Throw Away Those Overlapping Low Snr Bands
2022 12th Workshop on Hyperspectral Imaging and Signal Processing: Evolution in Remote Sensing (WHISPERS)
Final program and proceedings, Nov 1, 2021
Multispectral images contain more spectral information of the scene objects compared to color ima... more Multispectral images contain more spectral information of the scene objects compared to color images. The captured information of the scene reflectance is affected by several capture conditions, of which the scene illuminant is dominant. In this work, we implemented an imaging pipeline for a spectral filter array camera, where the focus is the estimation of the scene reflectances when the scene illuminant is unknown. We simulate three scenarios for reflectance estimation from multispectral images, and we evaluate the estimation accuracy on real captured data. We evaluate two camera model-based reflectance estimation methods that use a Wiener filter, and two other linear regression models for reflectance estimation that do not require an image formation model of the camera. Regarding the model-based approaches, we propose to use an estimate for the illuminant's spectral power distribution. The results show that our proposed approach stabilizes and marginally improves the estimation accuracy over the method that estimates the illuminant in the sensor space only. The results also provide a comparison of reflectance estimation using common approaches that are suited for different realistic scenarios.
Lecture Notes in Computer Science, 2016
One of the major issues related to dehazing methods (single or multiple image based) evaluation i... more One of the major issues related to dehazing methods (single or multiple image based) evaluation is the absence of the haze-free image (ground-truth). This is also a problem when it concerns the validation of Koschmieder model or its subsequent dehazing methods. To overcome this problem, we created a database called CHIC (Color Hazy Image for Comparison), consisting of two scenes in controlled environment. In addition to the haze-free image, we provide 9 images of different fog densities. Moreover, for each scene, we provide a number of parameters such as local scene depth, distance from the camera of known objects such as Macbeth Color Checkers, their radiance, and the haze level through transmittance. All of these features allow the possibility to evaluate and compare between dehazing methods by using full-reference image quality metrics regarding the haze-free image, and also to evaluate the accuracy of the Koschmieder hazy image formation model.
Lecture Notes in Computer Science, 2015
Single-sensor colour imaging systems mostly employ a colour filter array (CFA). This enables the ... more Single-sensor colour imaging systems mostly employ a colour filter array (CFA). This enables the acquisition of a colour image by a single sensor at one exposure at the cost of reduced spatial resolution. The idea of CFA fit itself well with multispectral purposes by incorporating more than three types of filters into the array which results in multispectral filter array (MSFA). In comparison with a CFA, an MSFA trades spatial resolution for spectral resolution. A simulation was performed to evaluate the colorimetric performance of such CFA/MSFA imaging systems and investigate the trade-off between spatial resolution and spectral resolution by comparing CFA and MSFA systems utilising various filter characteristics and demosaicking methods including intraand inter-channel bilinear interpolation as well as discrete wavelet transformed based techniques. In general, 4-band and 8-band MSFAs provide better or comparable performance than the CFA setup in terms of CIEDE2000 and S-CIELAB colour difference. This indicates that MSFA would be favourable for colorimetric purposes.
Applied sciences, Jun 29, 2022
This work investigates the use of Reflectance Transformation Imaging (RTI) rendering for visual i... more This work investigates the use of Reflectance Transformation Imaging (RTI) rendering for visual inspection. This imaging technique is being used more and more often for the inspection of the visual quality of manufactured surfaces. It allows reconstructing a dynamic virtual rendering of a surface from the acquisition of a sequence of images where only the illumination direction varies. We investigate, through psychometric experimentation, the influence of different essential parameters in the RTI approach, including modeling methods, the number of lighting positions and the measurement scale. In addition, to include the dynamic aspect of perception mechanisms in the methodology, the psychometric experiments are based on a design of experiments approach and conducted on reconstructed visual rendering videos. The proposed methodology is applied to different industrial surfaces. The results show that the RTI approach can be a relevant tool for computer-aided visual inspection. The proposed methodology makes it possible to objectively quantify the influence of RTI acquisition and processing factors on the perception of visual properties, and the results obtained show that their impact in terms of visual perception can be significant.
HAL is a multi-disciplinary open access archive for the deposit and dissemination of scientific r... more HAL is a multi-disciplinary open access archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers. L'archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d'enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
Linear minimum mean square error can be used to demosaic images from a colour-polarization filter... more Linear minimum mean square error can be used to demosaic images from a colour-polarization filter array sensor. However, the role of training data on its performance is yet an open question. We study the model selection using crossvalidation techniques. The results show that the training model converges quickly, and that there is no significant difference in training the model with more than 12 images of approximately 1.5 megapixels. We also found that the selected trained model performs better compared to a dedicated Colour-Polarization Filter Array demosaicing algorithm in terms of Peak Signalto-Noise Ratio.
Uploads
Papers by Jean-baptiste Thomas