You are here

Michela Lecca

Permanent researcher
  • Phone: 0461314536
  • FBK Povo
Short bio

Michela received her Master Degree in Mathematics from the University of Trento, Italy. She is currently researcher at the Research Unit Technologies of Vision of Fondazione Bruno Kessler of Trento, Italy. Her research interests include automatic object recognition, image retrieval, semantic image labeling, color constancy, color correction, and low-level image processing. From 2014, Michela collaborates with the research Unit IRIS (FBK-CMM) on hardware-oriented computer vision topics.

Michela took part to the program committee of many editions of Applied Computing Machine Symposia, and carries on a reviewing activity for several international conferences. She is a member of the International Association for Pattern Recognition -  Associazione Italiana per la ricerca in Computer Vision, Pattern recognition e machine Learning IAPR - CVPL (ex GIRPR) and of the Gruppo Italiano del Colore - Associazione Italiana Colore (GdC-AIC).

In 2015-2016 Michela led a Research in Pairs with the University of Trento, Department of Mathematics, and the University of Milano, Department of Computer Science. "Research in Pairs" is a program of FBK-CIRM, that promotes the collaboration among different research centers and/or university on topics mainly related to Mathematics. The research in pairs involving Michela focuses on the development of variational models for estimating the human color sensation. The main findings of this research have been published on JOSA A  and on IEEE TIP

In March 2017, Michela presented a Tutorial on Retinex theory at the 6th Computational Color Imaging Workshop (Milano, Italy) .

In September 2017, Michela, along with her colleagues at IRIS, participated at the Notte dei Ricercatori and presented some outcomes of the EU Project FORENSOR (Visione Intelligente a Basso Consumo / Ecco la telecamera anti-crimine!).

Michela was a co-organizer of the Workshop "Mathematics for Computer Vision" (MCV 2018), hosted by FBK - ICT on February 15-16, 2018.

Michela is a Guest Editor of the Special Issue on J. of Imaging Image Enhancement, Modeling and Visualization. The call for paper is published on the journal website, deadline on August 12, 2018 (EXTENDED!).

Research interests
object recognition semantic image labeling color constancy color correction hardware oriented image processing
  1. Celebi, Emre M.; Lecca, Michela; Smolka, Bogdan (eds.),
    Celebi, Emre M.; Lecca, Michela; Smolka, Bogdan,
  2. Michela Lecca,
    On the von Kries Model: Estimation, Dependence on Light and Device, and Application,
    Advances in Low-Level Color Image Processing,
    Celebi, M. Emre; Smolka, Bogdan,
    , pp. 95 -
  3. Michela Lecca,
    Color improves Texture Retrieval,
    Proc. of X Colour Conference,
    , (X Colour Conference,
    Genova - Italia,
    11-12 Sept. 2014)
  4. Michela Lecca ; Leonardo Gasparini ; Massimo Gottardi,
    Ultra-low power high-dynamic range color pixel embedding RGB to rg chromaticity transformation,
    Optical Sensing and Detection III,
    , (SPIE Photonics Europe,
    Brussels - Belgium,
    14 - 17 April 2014)
  5. P. Lecca; M. Lecca,
    Mechanistic Models of Astrocytic Glucose Metabolism Calibrated on PET Images,
    Biomechanics of Cells and Tissues,
    , pp. 131 -
  6. Michela Lecca,
    Methods for Estimating the von Kries Transform: A Review and A Comparison,
    Proc. of IX Color Conference,
    , (IX Color Conference,
    Firenze, Italia,
    Sept. 19-20, 2013)
  7. Michela Lecca; Mauro Dalla Mura,
    Intensity and Affine Invariant Statistic-based Image Matching,
    Computational Modeling of Objects Presented in Images,
    CRC Press
    Taylor & Francis Group
    , pp. 85-
    , (CompImage 2012 - Computational Modeling of Objects Presented in Images: Fundamentals, Methods, and Applications,
    Rome - Italy,
    09/05/2012 - 09/07/2012)
  8. Michela Lecca,
    Histogram-based Estimation of Illuminant Changes: von Kries Model versus Diagonal-Offset Model,
    Technical Report


    A same scene captured under different lights produces two different color images. Such a color variation is usually approximated by the von Kries model, that relates the color responses of the images by a diagonal linear map. In order to describe also image noise, diffuse lights possibly disturbing the scene or lens scattering, an offset term is often added.
    In this work, we compare the diagonal model and the diagonal-offset model in terms of accuracy on color correction and illuminant invariant image retrieval. Our experiments show just slight differences between the performances of the two models, that provide satisfactory results.

  9. Michela Lecca,
    Geometric Distortions of Color Channels for Linear Color Correction Between Images,
    Technical Report

    Abstract - A change of color between two pictures of a same scene captured by different devices and/or under different illuminants, is usually approximated by a full 3x3 linear map between the RGB responses of the images. By assuming this model, we develop a novel efficient color correction algorithm, with linear complexity in the number of image pixels. The main idea underlying our approach is to describe the images to be corrected by nine channels: the red, green, blue color channels provided by the device, plus six other channels encoding both geometric and color information. These additional channels are built up from the color ones by distorting them with 2D functions related to the pixel positions. The color correction is achieved by a linear transform which maps the mean values of the nine channels of first image on the mean values of the nine channels of the second images.

  10. M. Lecca; S. Messelodi,
    Linking the von Kries Model to Wien's Law for the estimation of an Illuminant Invariant Image,
    vol. 32,
    n. 15,
    , pp. 2086 -