Study of the model for estimating the degree of influence of distortion on image quality with the application of an entropic measure


The article deals with the problem of comparative analysis of the experimental results of using standard-free quality assessment measures depending on the type of distortion applied to the image. The previously developed measure based on the Weibullian model of the gradient magnitude and the entropy measures applied both to the original image and to the gradient characteristics of the ima ge are considered as No-Reference quality measures. Weibull distribution parameters are estimating using a set of image gradient magnitudes using the Sobel operator. A series of experiments was carried out on texture images from the well-known Brodatz texture database and the TID2013 database, which contains images distorted by various types. It was observed that texture images are in good agreement with the Weibullian model, which makes it possible to effectively apply the above-mentioned No-Reference measures and carry out a comparative analysis. The use of the TID2013 database with 3000 images, distorted by 24 types, each at five levels, makes it possible to evaluate additionally the effectiveness of the entropy approach to assessing image quality, comparing them with visual assessments also given in the database. Then it was illustrated that applying the entropy approach directly to the original image gives a worse result than applying image gradients to the magnitude. At the same time, a higher sensitivity to structural changes from the level and type of applied distortions is observed. The problem of the expediency of calculating the entropy using estimates of the parameters of the Weibull distribut ion instead of the histogram of the array of gradient magnitudes is investigated separately. The paper contains numerical and graphical materials illustrating the obtained results.



  1. V. A. Soyfer (Ed.), Methods of computer image processing; (in russian), Physmatlit, (2003).
  2. N. Ponomarenko, L. Jin, O. Ieremeiev, V. Lukin, K., Egiazarian, J. Astola, B. Vozel, K. Chehdi, M. Carli, F. Battisti, et al., Signal Proc.: Image Communication, 30, 57–77, (2015).
  3. H. Sheikh, Z. Wang, L. Cormack, A. Bovik, .
  4. D. M. Chandler, ISRN Signal Process, 1–53, (2013).
  5. Z. Wang, A. C. Bovik, IEEE Signal Proc. Letters, 9(3), 81–84, (2002).
  6. Z. Wang, A.C. Bovik, H.R. Sheikh, E.P. Simoncelli, IEEE Trans. Image Process., 13(4), pp. 600– 612, (2004).
  7. S. S. Hemami, A. R. Reibman, Signal Process. Image Commun, 25(7), 469–481, (2010).
  8. V. Kamble, K. M. Bhurchandi, Optik – Int. J. Light and Electron Optics, 126 (11-12),1090-1097, (2015).
  9. E. Cohen, Y. Yitzhaky, Signal, Image and Video Proc., 4(3), 289–302, (2010).
  10. V. V. Starovoitov, F. V. Starovoitov, Systems Analysis and Applied Informatics 1, 24-32 (in russian), (2017).
  11. D. Аsatryan, Computer Optics, 41(6), 957-962, (2017).
  12. D. Аsatryan, in: Int. Conf. CSIT, Yerevan, Armenia, 201–203, (2019).
  13. F. Escolano, P. Suau, B. Bonev, Information Theory in Computer Vision and Pattern Recognition. Springer-Verlag, (2009).
  14. D. Asatryan, K. Egiazarian, in: Int. Workshop Local and Non- Local Approximation in Image Processing, Finland, Helsinki, 70-73, (2009).
  15. D. G. Asatryan, Computer Optics; 43(2), 245-250, (2019).
  16. D. G. Asatryan, M. E. Haroutunian, Y. I. Golub, V. V. Starovoitov, System analysis and applied information science, 3, 22-27, (2020).

Article full text

Download PDF