Digital Image Interpolation in Matlab

3 Image Quality

The performance assessment of image interpolation algorithms can be categorized into objective and subjective assessments, and they are just the two faces of the mirror. Since the interpolated images are to be perceived by human eyes, therefore, subjective analysis is considered to be the final quality assessment of the interpolated image. However, one's medicine is the other's poison. It is difficult if not impossible to provide a subjective analysis to the interpolated image as it requires time and money and is highly inconvenient. Not to mention that there is no commonly accepted subject quality measure or feature sets for all varieties of image interpolation problems. Researchers are devoting massive efforts in developing different objective quality assessment algorithms that take the human vision system (HVS) into consideration (to model and to approximate the behavior of human vision) such as to provide an objective mean to compare the visible artifacts generated throughout the interpolation process. These algorithms give objective quality score that mimic the subjective quality measure for the image under test, without going through the subjective quality analysis. The objective scores (which are sometimes referred to as index) of different quality assessment algorithms depend on how the visible artifacts are quantified and also the sources of the reference data for comparison. Therefore, it is important for the readers to understand the definition ...