Using Set Partitioning in Hierarchical Trees in the EZW Algorithm Yeon Hyang Kim William...

6
Using Set Partitioning Using Set Partitioning in Hierarchical Trees in Hierarchical Trees in the EZW Algorithm in the EZW Algorithm Yeon Hyang Kim Yeon Hyang Kim William L’Huillier William L’Huillier

Transcript of Using Set Partitioning in Hierarchical Trees in the EZW Algorithm Yeon Hyang Kim William...

Page 1: Using Set Partitioning in Hierarchical Trees in the EZW Algorithm Yeon Hyang Kim William L’Huillier.

Using Set Partitioning in Using Set Partitioning in Hierarchical Trees in the Hierarchical Trees in the

EZW AlgorithmEZW Algorithm

Yeon Hyang KimYeon Hyang Kim

William L’HuillierWilliam L’Huillier

Page 2: Using Set Partitioning in Hierarchical Trees in the EZW Algorithm Yeon Hyang Kim William L’Huillier.

Introduction to wavelet Introduction to wavelet transformtransform

May contain multiple levelsMay contain multiple levels

Pixel magnitude levels decrease in 3 Pixel magnitude levels decrease in 3 quadrantsquadrants

Total size remains constantTotal size remains constant

Page 3: Using Set Partitioning in Hierarchical Trees in the EZW Algorithm Yeon Hyang Kim William L’Huillier.

Original idea of encodingOriginal idea of encoding

Use sorting passes to characterize Use sorting passes to characterize significance of information in bit depthsignificance of information in bit depth

Output the pixel coordinates that Output the pixel coordinates that corresponds to given significance levelscorresponds to given significance levels

Repeat in decrementing significance levelsRepeat in decrementing significance levels Much resources spend on sorting passMuch resources spend on sorting pass

Page 4: Using Set Partitioning in Hierarchical Trees in the EZW Algorithm Yeon Hyang Kim William L’Huillier.

Set Partitioning in Hierarchical Set Partitioning in Hierarchical TreesTrees

Elements are tagged significant or Elements are tagged significant or insignificant with respect to a given thresholdinsignificant with respect to a given threshold

Parse through the image in a tree formatParse through the image in a tree format Assumed information: root of tree has Assumed information: root of tree has

greatest informationgreatest information

Page 5: Using Set Partitioning in Hierarchical Trees in the EZW Algorithm Yeon Hyang Kim William L’Huillier.

Original image

Reconstruction from compression level 4

Rates: .2288 and .2177

Reconstruction from compression level 6

Rates: .1283 and .1332

Page 6: Using Set Partitioning in Hierarchical Trees in the EZW Algorithm Yeon Hyang Kim William L’Huillier.

implicationsimplications

This algorithm is able to send the This algorithm is able to send the most significant information first most significant information first before ending the transmitionbefore ending the transmition

JPEG2000JPEG2000

Questions?Questions?