Software & Datasets

Software

  • HDR full-reference quality metrics (2019).  We propose an HDR quality metric able to assess luminance and chromatic distortions caused by image compression.To obtain such a metric, we combine two full-reference quality metrics and two color image features using a Support Vector Regressor.
    • This metric is presented in the following paper:
      Rousselot, M., Le Meur, O., Cozot, R., & Ducloux, X. (2019, September). Quality metric aggregation for HDR/WCG images. In Image Processing (ICIP), 2019 IEEE International Conference on . IEEE.
    • Software (Matlab binary): Here
  • Perceptual metric for color transfer methods (2017). We propose a perceptual model for evaluating results from color transfer methods. We conduct a user study, which provides a set of subjective scores for triples of input, target and result images. Then, for each triple, we compute a number of image features, which objectively characterize a color transfer. To describe the relationship between these features and the subjective scores, we build a regression model with random forests. An analysis and a cross-validation show that the predictions of our model are highly accurate.
    • Details as well as the software are given : Here
    • This metric is presented in the following paper:
      Hristova, H., Le Meur, O., Cozot, R., & Bouatouch, K. (2017, September). Perceptual metric for color transfer methods. In Image Processing (ICIP), 2017 IEEE International Conference on (pp. 1237-1241). IEEE.

Datasets

Eye-tracking dataset

  • From the paper: Follet, B., Le Meur, O., & Baccino, T. (2011). New insights into ambient and focal visual fixations using an automatic classification algorithm. i-Perception, 2(6), 592-610.
    120 images, 800×600, belonging to 4 categories (Street, Coast, Mountain, and Open Country). Download link: HERE

HDR sequences datasets (3 datasets)

  • Desert – HDR sequences

    • Computer generated HDR sequences with sudden change in brightness 300 (1920×1080) frames openEXR format
    • Graded for 4.000 nits display
    • Free to non commercial use
    • Download : please contact Rémi Cozot

 

  • HDR sequences and subjective quality scores

    This dataset is composed of 8 reference images (without degradation) and 96 distorted images (Gaussian noise, Gamut mismatch, HEVC compression). Subjective scores for a panel of naive and expert observers are available.This dataset is presented in the following paper:Rousselot, M., Auffret E., Ducloux X., Le Meur O. and Cozot R., Impacts of viewing conditions on HDR-VDP2, in 26th European Signal Precessing Conference (EUSIPCO), Roma, Italy, Sept. 2018
    Download link: HERE
    This research is the result of a collaboration with Harmonic Inc.
  • 4Kdtb – High resolution HDR images and subjective scores

    This dataset is composed of 8  reference images and 96 distorted images (HEVC compression). This dataset is made available under the licence: CC-BY (more information: https://creativecommons.org/licenses/by/2.0/).The dataset is presented in the following paper:Rousselot, Maxime, Le Meur, Olivier, Cozot, Rémi, Ducloux, Xavier, Quality Assessment of HDR/WCG Images Using HDR Uniform Color Spaces, Journal of Imaging, 2019, vol. 5, no 1, p. 18.
    Download link: HERE (>700Mo)
    This research is the result of a collaboration with Harmonic Inc.

Comments are closed.