Finally, blurring is related to aliasing in the sense that it is also a mismatch between an intermediate data representation and their final downsampled or oversampled version. In this case, the mismatch is such that the intermediate data is too coarse for the task. This results in an image that appears to be out of focus. When the filter associated to the synthesis function ^ or ^int is very different from an ideal filter, aliasing and blurring can occur simultaneously (they usually do). Note that, both for aliasing and blurring considerations, the intermediate representation need not be explicitly available or computed. To highlight blurring, it is enough to iterate the same interpolation operation several times, thus effectively magnifying the effect. Figure 4 has been obtained by the compound rotation of an image by 36 steps of 10° each.

FIGURE 2 Aliasing. (Top) Low quality introduces a lot of aliasing. (Bottom) Better quality results in less aliasing. (Both) At too coarse scale, the structural appearance of the bundles of cells is lost.

FIGURE 3 Blocking. After low-quality magnification, the highlighted area (left) appears pixellated (center). Better-quality magnification results in less pixellation (right).

Was this article helpful?

0 0

Post a comment