Study with the several resources on Docsity
Earn points by helping other students or get them with a premium plan
Prepare for your exams
Study with the several resources on Docsity
Earn points to download
Earn points by helping other students or get them with a premium plan
computer graphics - compression topic
Typology: Exams
1 / 59
This page cannot be seen from the preview
Don't miss anything!
Audio, image and video require vast
amounts of data
Low networkās bandwidth doesn't allow for
real time video transmission
Slow storage or processing devices don't
allow for fast playing back
Compression reduces storage requirements
ļ± Lossless: recover the original
representation
ļ± Lossy: recover a representation
similar to the original one
ļ± high compression ratios
ļ± more practical use
ļ± Hybrid: JPEG, MPEG, px64 combine
several approaches
Furht at.al. 96
Furht at.al. 96
JPEG Block Diagrams
ļ± Three main blocks:
ļ± Forward Discrete Cosine Transform (FDCT)
ļ± Quantizer
ļ± Entropy Encoder
ļ± Essentially the sequential JPEG encoder
ļ± Main component of progressive, lossless
and hierarchical encoders
ļ± For gray level and color images
ļ±
F(0,0) is the DC coefficient: average
value over the 64 samples
ļ± The remaining 63 coefficients are the
AC coefficients
ļ± Pixels in [-128,127]: DCTs in
[-1024,1023]
ļ± Most frequencies have 0 or near to 0
values and need not to be encoded
ļ± This fact achieves compression
ļ± All 64 DCT coefficients are quantized
ļ± F
q
(u,v) = Round[F(u,v)/Q(u,v)]
ļ± Reduces the amplitude of coefficients
which contribute little or nothing to 0
ļ± Discards information which is not
visually significant
ļ± Quantization coefficients Q(u,v) are
specified by quantization tables
ļ± A set of 4 tables are specified by JPEG
Furht at.al. 96
ļ± Predictive coding of DC Coefficients
ļ± Adjacent blocks have similar DC intensities
ļ± Coding differences yields high
compression
ļ± Converts each sequence into binary
ļ± First DC following with ACs
ļ± Huffman tables are specified in JPEG
ļ±
Each (runlength, size) is encoded
using Huffman coding
ļ± Each (amplitude) is encoded using a
variable length integer code
ļ± (1,4)( 12 ) => (1111110110 1100 )
Compression ratio (CR): increases with
higher compression
Root Mean Square Error (RMS): better
quality with lower RMS
i
i
ļ½
n
i
i i
1
2
