Human perception of geometric distortions in images

Abstract

We present in this paper the results of our study on the human perception of geometric distortions in images. The ultimate goal of this study is to devise an objective measurement scheme for geometric distortions in images, which should have a good correspondence to human perception of the distortions. The study is divided into two parts. The first part of the study is the design and implementation of a user-test to measure human perception of geometric distortions in images. The result of this test is then used as a basis to evaluate the performance of the second part of the study, namely the objective quality measurement scheme. Our experiment shows that our objective quality measurement has good correspondence to the result of the user test and performs much better than a PSNR measurement.

A Comparative Perceptual Study of Soft-Shadow Algorithms

“…Note that the contrast of enlarged images was adjusted to better visualize differences. et al 2008], tone-mapping [Ledda et al 2005], geometric distortions [Setyawan and Lagendijk 2004], image-retargeting algorithms [Rubinstein et al 2010], and global illumination [Čadík et al 2012]. The structure of our study is naturally related to these, and we discuss similarities as well as differences with respect to them in the following.…”

“…To analyze the quality of the results in terms of consistence within comparison responses and agreement between different participants, we compute the so-called coefficient of consistence and coefficient of agreement [Kendall and Gibbons 1990], as proposed by Setyawan and Lagendijk [2004]. Since we are also interested in the informative value given by the amount of neutral responses, we define a new measure which we denote as degree of indifference.…”

“…To obtain an estimate of similarity in our study as well, we incorporate the number of neutral responses into the evaluation. Setyawan and Lagendijk [2004] analyze the quality of responses of single participants and the agreement between all participants. We adopt this concept in our study to evaluate the performance of participants, and whether the learning phase has been effective.…”

“…The coefficient of consistence ξ is a measure to score the amount of contradictory responses. It is computed by counting or estimating the amount of so-called circular triads which occur when algorithm i is preferred over j, j over k, and k over i (see [Setyawan and Lagendijk 2004] …”

“…Aij 2 which agree in their choice [Setyawan and Lagendijk 2004]. It is a measure to evaluate the ability of participants to make "good" judgments in the comparison task (e.g., to identify the more physically correct shadow).…”