Abstract

The problem of blindly assessing the quality of visual signals – without reference, and without assuming a single distortion type – has long been regarded as, if not impossible, perhaps too difficult to bother with. After all, it requires sorting out the black box we call the visual apparatus of the human brain, which despite an increased level of transparency over the past 40 years, remains poorly understood. Moreover, it involves the very un-engineering-like concept of subjectivity, an uncomfortable thing altogether for the analytic types populating our profession. And lastly, it requires dispensing with older ideas of quality such as fidelity, similarity, and metric comparison. In this talk I will discuss our recent efforts on blind or “no reference” image quality assessment problems, including machine learning approaches, and the looming question of stereo (3D) image quality.

© 2010 Optical Society of America