Skip to Main Content
While visual models have been used for several decades to improve the perceived image quality of image processing algorithms and display design, the complexity of these visual models has been steadily increasing as progress is made in basic visual science. Beginning in the early 1990's applied visual models began to use not only a multiband approach in radial spatial frequency (e.g., the Laplacian pyramid) but also in orientation. At this point, the ability to predict the important effects of masking became more accurate. This paper will provide a close look at the role of these orientation bands in the application image scaling, which must balance the tradeoffs between sharpness, aliasing, and ringing, all acting in concert with the masking caused by the signal itself.