In the paper an Internet based massive experiment devoted to subjective assessment of images quality is described. The investigations have been based on DSIS method with some modifications induced by the nature of Internet poll. Special measures have been undertaken for removing results given by incompetent experiment participants. The huge number of tests allowed for multi-aspect image quality evaluation: 12 deformation effects have been applied to 22 images. The results point at important problems linked with automation of image quality assessment: factors determined by high levels of human visual cognition (e.g. face deformation), surprisingly high impact of local distortions on scores for some tested images, finally, correct modeling of low-level effects, like masking.
[1]
Sugato Chakravarty,et al.
Methodology for the subjective assessment of the quality of television pictures
,
1995
.
[2]
Hua Li,et al.
Image Quality Assessment Using Rough Fuzzy Integrals
,
2007,
27th International Conference on Distributed Computing Systems Workshops - Supplements (ICDCSW'07).
[3]
Ahmet M. Eskicioglu,et al.
Comparison of three full-reference color image quality measures
,
2006,
Electronic Imaging.
[4]
Alan C. Bovik,et al.
Image information and visual quality
,
2004,
2004 IEEE International Conference on Acoustics, Speech, and Signal Processing.