Computational Image Quality

Images are a powerful, efficient means for communicating information, spurring advances in technologies underlying image capture, transfer, storage and display. This text asks the fundamental question: what is image quality? The answer requires that we think of images not as signals, but as carriers of visual information, and the visuo-cognitive processing of images as information processing. This processing is an essential stage in human interaction with the environment; the perceived adequacy of an image is based on comparisons to our memory standards of what is natural and correct. With this in mind, the author presents partially flexible metrics and methods for predicting the usefulness and naturalness of reproductions, including the measurement of discriminability and identifiability. Algorithms that summarize these concepts complete this analysis.