|
Abstract:
Wavelets have been around for 20 years and are now included among the
standard techniques for lossy image compression, whose objective is to
transform information in such a way that irrelevant information can be
discarded while relevant information is (mostly)
retained. Historically, singular value decompositions (SVDs) and Fast
Fourier Transforms (FFTs) were used for this. The talk will contain an
introduction to wavelets and their properties, and some comments about
the newer mathematical understanding of how wavelets capture "relevant
information" in images. Comparisons between wavelets and other
methods will be demonstrated on specific images such as one of my dog
Max, and further evolution of the underlying mathematics will be outlined.
|