Image Compression: Comparative Analysis of Basic Algorithms

1
Image Compression Comparative Analysis of Basic
Algorithms

• Yevgeniya Sulema (Ukraine)
• Samira Ebrahimi Kahou (Iran)
• National Technical University of Ukraine
• Kyiv Polytechnic Institute
• sulema_at_scs.ntu-kpi.kiev.ua
• samira_ebrahimi_at_hotmail.com

2
Outline

• Existing compression methods and classification
• Criteria
• How to choose image set for testing
• Realizing algorithms
• Getting numerical values on chosen criteria
• Verifying results obtained from test
• Analysis and conclusion

3
Compression algorithms-Classification

• 5 Main Classification Types chosen.
• By data type
• General algorithms
• Algorithms for audio-compression
• Algorithms for image-compression
• Algorithms for video-compression

4
Compression algorithms-Classification (..2)

• By data source
• Dynamic
• Static
• By redundancy type
• Statistical redundancy reduction
• Spatial redundancy reduction

5
Compression algorithms-Classification (..3)

• By restoring the original dataset
• Lossless
• Lossy
• By computational approach
• Statistical
• Dictionary
• Transformation based
• Hybrid

6
Classes of Images

• Business graphics (schemes, diagrams, charts)
• Pictures created in graphic editors (photoshop)
• Photorealistic images (photos, textures)
• Coefficient of correlation can be used between an
  analyzed (test) image and an etalon image to
  classify images

7
Sample images

• Image with two monochrome areas
• Image with large monochrome fields
• Gradient image
• Image with small monochrome fields

8
Correlation coefficients(Sample images)

9
Criteria

• Compression ratio
• Time of compression
• Time of decompression
• Peak signal-to-noise ratio (PSNR)
  MSE Mean Squared Error
• Coefficient of correlation between original and
  decompressed image

10
Matlab image processing Toolbox

11
Why Matlab?

• It provides a comprehensive set of
  reference-standard algorithms.
• The software is a collection of functions that
  extend the capability of the MATLAB.
• The toolbox supports a wide range of image
  processing operations.
• Most toolbox functions are written in the open
  MATLAB language, giving us the ability to inspect
  the algorithms, modify the source code.

12
Algorithms

• Lossless
• LZW
• LZ77
• Huffman
• Adaptive Huffman
• Shannon-Fano
• Arithmetic

• Lossy
• JPEG(Coarse and Fine)
• Wavelet(Daubechies, Coiflets, Symlets, Discrete
  Meyer wavelet, Biorthogonal, Reverse
  Biorthogonal)
• SPIHT
• Fractal

13
LosslessTime of compression

14
LosslessTime of decompression

15
LosslessCompression ratio

16
Lossless Algorithm Observation

• Dictionary Based Algorithms most Effective
• LZ77 prime example from our research
• Minimal Time for Compression
• Minimal Time for Decompression
• High Compression Ratio

17
LossyTime of compression

18
LossyTime of decompression

19
LossyCompression ratio

20
LossyCorrelation coefficient

21
LossyPSNR

22
Lossy Algorithm Observations

• Fractal Algorithm not practical.
• All remaining algorithms are Hybrid
• Combination of procedures can result in increased
  quality.

23
Conclusion

• Our research allows us to draw 3 main
  conclusions
• The selection of the proper compression algorithm
  for each image class should be made
• Hybrid algorithms, JPEG, can be modified in order
  to achieve better result
• Combination of a dictionary and transforms most
  promising.

24
Thank YOU!

• Questions??? samira_ebrahimi_at_hotmail.com