ABSTRACT
In the preceding chapter, a psychocentric viewpoint was proposed, in which the existence of orthogonal perceptual attributes combining according to the multivariate formalism is assumed. Even if this viewpoint is valid, however, there is no guarantee that these orthogonal attributes can easily be identified, nor that correlating objective metrics can be readily designed. In nearly all our studies, identification of orthogonal image quality attributes has been reasonably straightforward and design of predictive objective metrics, while often challenging, has ultimately proven feasible. In a few instances, however, a fully satisfactory treatment was not developed during the initial investigation. In this chapter, the most notable example of this type is described so that the reader can appreciate some of the difficulties that may be encountered in practice.
