Scribd
Upload
87 views6 pages

Wavelet Based Image Compression Using SPIHT Algorithm: Abstract

This document provides an overview of wavelet based image compression using the Set Partitioning In Hierarchical Trees (SPIHT) algorithm. It describes how SPIHT encodes an input image by first segmenting it and applying sub-band transforms, then encoding the transformed image using SPIHT coding. At the decoder, the coded image is decoded and an inverse transform is applied to reconstruct the original image. SPIHT improves upon previous algorithms by scanning wavelet coefficients along quadtrees instead of sub-bands and separately encoding the signs and magnitudes of coefficients.

Uploaded by

yrikki
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOC, PDF, TXT or read online on Scribd
87 views6 pages

Wavelet Based Image Compression Using SPIHT Algorithm: Abstract

This document provides an overview of wavelet based image compression using the Set Partitioning In Hierarchical Trees (SPIHT) algorithm. It describes how SPIHT encodes an input image by first segmenting it and applying sub-band transforms, then encoding the transformed image using SPIHT coding. At the decoder, the coded image is decoded and an inverse transform is applied to reconstruct the original image. SPIHT improves upon previous algorithms by scanning wavelet coefficients along quadtrees instead of sub-bands and separately encoding the signs and magnitudes of coefficients.

Uploaded by

yrikki
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOC, PDF, TXT or read online on Scribd
You are on page 1/ 6

A Project Synopsis On

Wavelet Based Image Compression Using SPIHT Algorithm

Abstract:
In this wor we have st!died well nown state o" the art image compression algorithms# These codecs are $ased on wavelet trans"orms in most cases# Their compression e""iciency "or e#g on image compression is widely ac nowledged# The already !sed %P&'())) standard* e#g#* will $e $ased on wavelet trans"orms too# We present the well nown algorithm o" Pearlman Set Partitioning In Hierarchical Trees SPIHT to restrict the necessity o" random access to the whole image to a small s!$ images only# The main idea is $ased on partitioning o" sets* which consists o" coe""icients or representatives o" whole s!$ trees# The decoder d!plicates the e+ec!tion path o" the encoder to ens!re this $ehavior* the coder sends the res!lt o" a $inary decision to the decoder $e"ore a $ranch is ta en in the algorithm# Th!s* all decisions o" the decoder are $ased on the received $its#The name o" the algorithm is composed o" the words set and partitioning. The compression per"ormance is meas!red with pea $e implemented thro!gh ,AT-AB# signal to noise ratio

compared to the original codec remains still the same or nearly the same# This code can

(./

Block Diagram:
As shown in "ig 0a1 in SPIHT &ncoder* the in p!t image can $e s!$ divide into constit!ent parts or o$jects* as well as s!$ $and trans"orms can $e applied on segmented image i#e# image is s!$divided into several "re2!ency $ands and each $and is digitally encoded separately $y SPIHT &ncoding#

image

Segmentation

S!$ Band Trans"orm

SPIHT &ncoding

Coded image

Percept!al weights

4ig 0a15 SPIHT &ncoder

As shown in "ig 0a1 in SPIHT 3ecoder* the coded image can $e decoded as well as inverse trans"orm can $e applied on it# So we can get the original reconstr!cted image#

Coded Image

&+tract Segmentation SPIHT 3ecoding

Inverse Trans"orm

9econstr!cted Image

Percept!al Weights

4ig 0$15 SPIHT 3ecoder 6./ SPIHT scans wavelet coe""icients along 2!ad 7tree instead o" s!$ $and# Another modi"ication is* SPIHT identi"ies signi"icance only $y the magnit!de o" wavelet coe""icients and encodes the sign separately into a new $it stream# Com$ined with adaptive arithmetic coding* SPIHT improved the compression per"ormance 2!ite m!ch and ca!sed a n!m$er o" new developments# The n!m$er o" $ytes can $e !sed to represent an image $y !sing wavelet compression 0&8W1 is less as compared to %P&' compression #By the !sage o" SPIHT algorithm*we can improve the compression per"ormance as compared to previo!s algorithms#

Algorithm Flow:

Initialization n : ;log( 0ma+ <coe""<1= -IP : All elements in H -SP : &mpty -IS : D>s o" 9oots Significance Map Encoding !Sorting "ass#$ "rocess %I" "or each coe"" 0i*j1 in -IP O!tp!t Sn0i*j1 I" Sn0i*j1:? O!tp!t sign o" coe""0i*j15 ).? : @.A ,ove 0i*j1 to the -SP &ndi" &nd loop over -IP B./

"rocess %IS "or each set 0i*j1 in -IS i" type D Send Sn0D0i*j11 I" Sn0D0i*j11:? "or each 0 *l1 o!tp!t Sn0 *l1 i" Sn0 *l1:?* then add 0 *l1 to the -SP and o!tp!t sign o" coe""5 ).? : @.A i" Sn0 *l1:)* then add 0 *l1 to the end o" the -IP end"or endi" else 0type % 1 Send Sn0%0i*j11 &0i*j1

I" Sn0%0i*j11:? add each 0 *l1 end i" on type &nd loop over -IS 'efinement "ass "rocess %S" "or each element 0i*j1 in -SP 7 except those j!st added a$ove O!tp!t the nth most signi"icant $it o" coe"" &nd loop over -SP (pdate 3ecrement n $y ? 'o to Signi"icance ,ap &ncoding Step Adaptive Arithmetic Code 0Optional1 &0i*j1 to the end o" the -IS as an entry o" type D remove 0i*j1 "rom the -IS

You might also like