Yann GAVET

  • Skills

    image processing, linux, augmented reality
    C/C++, python, matlab, OpenCL

  • Research Topics

    Modelisation and simulation of surfaces, mosaics and textures
    Voronoi diagrams are classically used to model mosaic patterns (like grains or cells). In the CorImMo 3D project (ANR TecSan 2012), the study of random fields allows us to randmly determine germs as well as distances to generate a spatial partition of the 2D or 3D space. The concepts and tools are based on:
    • random fields theory (covariance function, statistical moments including the skewness and kurtosis)
    • integral geometry (e.g., Minkowski functionals: area, perimeter and connectivity number for 2D fields; volume, surface area, mean breadth and connectivity number for 3D fields)
    • and mathematical morphology (e.g., openings and closures, skeletons, watersheds, granulometry).
    Mosaics
    The corneal endothelium is constituted of non-regenerative cells. We have proposed:
    • Segmentation methods based on human visual properties (Gestalt and Marr theories)
    • Segmentation evaluation tool, called epsilon dissimilarity criterion, tolerant to visual and spatial variations
    • 3D Reconstruction method (shape from focus), applied to corneal endothelium.
    CoLIP: Color Logarithmic Image Processing
    The Color Logarithmic Image Processing (CoLIP) is a recently introduced (2011) extension of the LIP mathematical framework based on abstract linear mathematics which provides a set of specific algebraic and functional operations that can be applied to the processing of color images valued in a bounded range. The CoLIP approach has been proved to be psycho-physically consistent with several laws and characteristics of the human color perception.
    Successful applications have been reported on the white balance correction in color images.

  • Teaching Areas

    Introduction to C programming, Signal Processing (L3)
    Introduction to image processing (M1), advanced image processing (M2), High Performance Computing (openmp, openCL).
    Taught in French and/or English

  • Biography

    His research is at the crossroad of mathematics (for the scientific background), physics (for the image formation rules), computer engineering (for testing and constructing the practical solutions) and the application fields.

  • Education

    Engineer, 2001
    Master thesis in image processing, 2004
    PhD thesis in image processing, 2008
    Habilitation 2015

  • Career History

    Assistant Professor
    Leader of the CorImMo3D project (ANR 2012, 650 K€)

    4 months (2017) detachment in Thalès-Angénieux, Saint-Heand, France.
    formerly: IT developer for CNES (Space French Agency)

53 documents

  • Yann Gavet, Mathieu Fernandes, Johan Debayle, Jean-Charles Pinoli. Dissimilarity criteria and their comparison for quantitative evaluation of image segmentation: application to human retina vessels. Machine Vision and Applications, Springer Verlag, 2014, 25 (8), pp.1953-1966. ⟨10.1007/s00138-014-0625-2⟩. ⟨hal-01094137⟩
  • Yann Gavet, Ola Ahmad, Jean-Charles Pinoli. Integral Geometry of Linearly Combined Gaussian and Student-t, and Skew Student's t Random Fields. First International Conference, Geometric Science of Information - GSI 2013 Proceedings, Aug 2013, Paris, France. pp.449-456, ⟨10.1007/978-3-642-40020-9_49⟩. ⟨hal-00870350⟩
  • Ola Ahmad, Yann Gavet, Jean Geringer, Jean-Charles Pinoli. Roughness variability estimation of microscopic surfaces during engineering wear process-Application to total hip implant. 11th International Conference on Quality Control by Artificial Vision, May 2013, Fukuoka, Japan. pp. ⟨hal-00870518⟩
  • Yann Gavet, Jean-Charles Pinoli. A geometric dissimilarity criterion between Jordan spatial mosaics. Theoretical aspects and application to segmentation evaluation. Journal of Mathematical Imaging and Vision, Springer Verlag, 2012, 42 (1), pp.25-49. ⟨10.1007/s10851-011-0272-4⟩. ⟨hal-00660178⟩
  • Yann Gavet, Mathieu Fernandes, Jean-Charles Pinoli. Quantitative evaluation of image registration techniques in the case of retinal images. Journal of Electronic Imaging, SPIE and IS&T, 2012, 21 (2), pp.21118. ⟨10.1117/1.JEI.21.2.021118⟩. ⟨hal-00700870⟩
  • Mathieu Fernandes, Yann Gavet, Jean-Charles Pinoli. Surface topography and texture restoration from sectional optical imaging by focus analysis. Intech. Image Restoration - Recent Advances and Applications, Aymeric Histace, pp.73-96, 2012, 978-953-51-0388-2. ⟨hal-00685364⟩
  • Mathieu Fernandes, Yann Gavet, Jean-Charles Pinoli. Robust 3-D reconstruction of surfaces from image focus by local cross-sectional multivariate statistical analyses: Application to human ex vivo corneal endotheliums.. Medical Image Analysis, Elsevier, 2012, 16 (6), pp.1293-1306. ⟨10.1016/j.media.2012.05.004⟩. ⟨hal-00733040⟩
  • Hélène Gouinaud, Yann Gavet, Johan Debayle, Jean-Charles Pinoli. Color Correction in the Framework of Color Logarithmic Image Processing. 7th International Symposium on Image and Signal Processing and Analysis (ISPA 2011), Sep 2011, Dubrovnik, Croatia. Non paginé. ⟨hal-00623449⟩
  • Yann Gavet, Mathieu Fernandes, Jean-Charles Pinoli. Automatic quantitative evaluation of image registration techniques with the "epsilon" dissimilarity criterion in the case of retinal images.. Tenth International Conference on Quality Control by Artificial Vision QVAC 2011, Jun 2011, Saint Etienne, France. pp.8000-23, ⟨10.1117/12.890883⟩. ⟨hal-00614076⟩
  • Mathieu Fernandes, Yann Gavet, Jean-Charles Pinoli. Improving focus measurements using logarithmic image processing. Journal of Microscopy, Wiley, 2011, 242 (3), pp.228-241. ⟨10.1111/j.1365-2818.2010.03461.x⟩. ⟨hal-00643563⟩