0

From https://github.com/plroit/Skyreach/wiki/Understanding-scalability-in-JPEG2000

an input image is divided into four images, each a quarter of the size of the input. Each of the four images is called a subband. One subband that is labeled 'LL' is actually a scaled-down version of the input, with a scaling factor of a 1/2.

If we take a single tiled image and do 2-D wavelet decomposition once, encode and send it.
Then decoder will find 4 sub bands.
Using that it will be able to draw max resolution picture of dimensions halved.
Isn't it?

Brian Tompsett - 汤莱恩
  • 5,753
  • 72
  • 57
  • 129
sumeet inani
  • 1
  • 4

1 Answers1

0

Nope, a decoder performs inverse discrete wavelet transformation to obtain an image of original dimension.

This transformation can be done lossily or losslessly.

The purpose of the transformation in an encoder is to separate information which can be compressed more effectively by arithmetic coder.

stuhlo
  • 1,479
  • 9
  • 17
  • ok . that means , those 4 sub bands are not directly related to image resolution. – sumeet inani May 05 '17 at 07:08
  • They are, the first level of the transformation produces 4 bands of half of the size of original image in each dimension. To obtain original image, you have to perform an inverse transformation which produces image of original size. One of the band, LL band, can be considered as a downscaled original image. – stuhlo May 05 '17 at 12:34