Image quality in context

An analysis of the ergonomic quality of the current standards for the visual display quality leads to a number of recommendations for the development of new international standards: – Separation for different types of users, esp. display designers, purchasers, and end users, -Independence of display technology to allow comparison, -Modular construction with several quality grades to allow benchmarking for different types of applications, -A test method for the end user standard that can be performed at the place of work, to take into account the effects of wear and drift of components and to be able to correct suboptimal configurations. The separate parameters that exert influence on the image quality of a broad category of images in the context of use, and their mutual coherence within the cycle of evaluation and adaptation of image quality are presented in the “Image Quality in Context Cycle”-model. The features that together form image quality are classified in a framework, which allows the image quality factors to be considered independent from the technology that formed the image. The construction in five different domains and eight types of features can be used for the systematic development of standards and image quality research. The practical problems of image quality maintenance are examined with regard to the correct and consistent display of colors. This aspect is also important for the average personal computer user, e.g. with the processing of digital photographs, the preparation of presentations and the judgment of possible online purchases. A number of substantial problems with existing models for the tone reproduction curve (TRC) of cathode ray tubes (CRT) and the methods to determine these curves by fitting from light measurements are discussed. The current power function models assume a constant exponent gamma and possible zero offsets for voltage and/or luminance, and are fitted based on minimizing the error in the physical luminance domain. A newly introduced method with an optimization in the standardized perceptual lightness domain produces better models with more accurate estimations of zero offsets. It also turns out that for some CRTs better results are reached if the assumption of a constant gamma is released. The practical possibilities of the user to optimize the color reproduction of the display by adjusting the “contrast” and “brightness” controls, and the visual measurement of monitor gammas are investigated in an experiment. 32 Subjects were asked to optimally set the black level according to two types of instruction (short and extended, between subjects under tow levels of illumination (dim and office, between subjects) for two types of display (CRT and Liquid Crystal Display (LCD), within subjects). Most of the subjects were not able to adjust the black to a level near the optimal setting, independent of ambient illumination, display or type of instruction. The technical constraints of the displays could have played a role here. The theoretical merits and practical problems of several existing methods for visual measurement of the monitor gamma are discussed and compared, and improvements are proposed. A new method to compare gamma models with a different number of parameters (with or without zero offsets) is introduced. In the experiment different methods of brightness matching are tested: spatial methods (a field with a striped pattern matched to a uniform field) with three different resolutions, and a temporal method (flickering field compared with a uniform field). The ambient illumination had no effect on the estimated gamma. Subjects had considerable problems with spatial brightness matching with low resolution. Temporal and spatial methods produced higher values for gamma for untrained subjects than for experts and estimations based on photometric measurements.