Re: pixel prediction for lossless image compression


Michael Schöberl wrote:

If you have images with per-pixel noise - what about a linear
combination of p6, p7, p8, p10 depending on you valuaA, valuaB valuaC ?

bye,
Michael

That sounds like a good idea.

I am also able two predict the abs(error) by using the smallest
value(valueA/valueB/valueC). After using some kind of sort methode I
was able to compress the image a bit more.

Do you have any experience in using subpixels for prediction.
Getting the subpixels by using some kind of interpolation.

.

Relevant Pages