Objective quality estimation for digital images and video streams

In the context of multimedia and hypermedia systems, the aspects of image quality are gaining increasingly in importance. Two tasks have to be fulfilled. First, a suitable quality level of a visual presentation for a specific task in the context of compression and bandwidth limitation has to be achieved. Second, an optimized design of multimedia applications for complete control of the quality of image sand video sequences is necessary. This paper provides a definition and discussion of the term quality with respect to the use of visual information in multimedia applications which leads to the development of a classification scheme for image quality aspects. Relationships between task-oriented presentations and quality aspects, quality parameters and quality control are presented. A short introduction into state-of-the-art image quality measurement is given and the advantages and limitations of the individual techniques are discussed. Based on the analysis of the measurement methods MT FA and SQRI, we present two enhancements which allow the application of this two measurements to digital systems. Thes modifications and enhancements trat the aspects of frequency amplifications and attenuation and the consideration of the human perception threshold. As an application of our classification scheme and the enhancements of MTFA and SQRI, we present approaches to estimate the quality of the color reproduction of the digital imaging system and to integrate several measurements into a general quality measurement environment (GQME). Finally, questions concerning the achieved results and future work and raised.