We explore the unifying connection between kernel regression, Volterra series expansion and multiscale signal decomposition using recent results on function estimation for system identification. We show that using any of these techniques for (non-linear) image processing tasks is (approximately) equivalent. Further, we use the relation between wavelets and independent components of natural images. Kernel methods can be shown to be implicit Volterra series expansions, which are well approximated by wavelets. Wavelets are, in turn, well represented by independent components of natural images. Thus, it can be seen that kernel methods are also near optimal in terms of higher order statistical modeling and approximation of (natural) images. This explains the reason for good results often (perceptually) observed with the use of kernel methods for many image processing problems.
Comments: 6 Pages.
[v1] 2018-04-19 09:47:02
Unique-IP document downloads: 20 times
Vixra.org is a pre-print repository rather than a journal. Articles hosted may not yet have been verified by peer-review and should be treated as preliminary. In particular, anything that appears to include financial or legal advice or proposed medical treatments should be treated with due caution. Vixra.org will not be responsible for any consequences of actions that result from any form of use of any documents on this website.
Add your own feedback and questions here:
You are equally welcome to be positive or negative about any paper but please be polite. If you are being critical you must mention at least one specific error, otherwise your comment will be deleted as unhelpful.