Image Resolution Enhancement

1
DIGITAL SIGNAL PROCESSINGIN ANALYSIS OF
BIOMEDICAL IMAGES
Prof. Aleš Procházka Institute of Chemical
Technology in Prague Department of Computing and
Control Engineering Digital Signal and Image
Processing Research Group
2
1. INTRODUCTION

• MOTIVATION OF THE DSP RESEARCH GROUP
• INTEGRATION ROLE OF SIGNAL
• AND IMAGE PROCESSING IN
• THE FRAME OF INFORMATION
• ENGINEERING
• Interdisciplinary area connecting
• mathematics and engineering
• control, measuring engineering, vision, speech
  processing,
• biomedicine, environmental engineering
• Fundament for data acquisition, system
  identification
• and modelling, signal de-noising, feature
  extraction,
• segmentation, classification, compression,
  prediction,
• Similar mathematical background based on methods
  of
• time-frequency and time-scale analysis in
  different areas

3
2. APPLICATIONS
INTERESTS OF DSP RESEARCH GROUP
Signal Prediction
Environmental Engineering
Biomedical Image Analysis
Remote Data Processing
4
3. TIME-FREQUENCY ANALYSIS
DISCRETE FOURIER TRANSFORM IN RESOLUTION
ENHANCEMENT
2-D DFT for k0,1,,N/2 1, l 0,1,,M/2
1 and f1(k)k/N , f2(l)l/M
1-D DFT for k0,1,,N/2 1 and f(k)k/N

5
4. TIME-SCALE ANALYSIS
WAVELET TRANSFORM IN SIGNAL PARTS DETECTION

• Initial wavelet defined either in the
• analytical form or by a dilation equation
• Dilation and translation
• coefficients a2m, bk 2m
• Initial wavelet is a pass-band filter
• Wavelet dilation corresponds
• to its pass-band compression

6
5. DENOISING OF SIGNAL / IMAGE COMPONENTS
WAVELET TRANSFORM IN IMAGE DENOISING

• ALGORITHM
• Decomposition stage convolution of a given
  signal and the filter
• 
  downsampling by D
• Coefficients
  - by rows and columns
• thresholding

Magnetic resonance image

• Reconstruction stage
• row upsampling by
• factor U and
• row convolution
• sum of the
• corresponding
• images
• column upsampling
• by factor U and
• column convolution

7
6. MR IMAGE RESOLUTION ENHANCEMENT
WAVELET TRANSFORM IN IMAGE RESOLUTION ENHANCEMENT
I. Image Resolution Enhancement using DFT

• MAGNETIC RESONANCE
• IMAGES OF A HUMAN BRAIN
• Original resolution
• 512 x 512 pixels
• Resolution enhancement
• 1024 x 1024 pixels

II. Image Resolution Enhancement using DWT

• CONCLUSIONS
• DFT the structures and
• edges are very smooth
• DWT sharper edges
• obtained

• DFT and DWT
• various methods to
• enhance the resolution
• can be applied

8
7. IMAGE RESTORATION
METHODS OF IMAGE COMPONENTS RESTORATION

• METHODS
• Detection of features of missing regions and
  their replacement by the
• most similar ones

• Multidirectional prediction of
• missing image
• parts
• Multidemensional cubic and spline interpolation
• Iterated wavelet interpolation

9
8. ITERATED WAVELET TRANSFORM IN IMAGE
RESTORATION
WAVELET TRANSFORM IN ITERATED INTERPOLATION

• ALGORITHM
• Image decomposition into a selected level
• Wavelet coefficients thresholding
• Image reconstruction
• Replacement of values outside regions of interest
  by original values
• The next iteration of image decomposition

10
9. IMAGE SEGMENTATION
WATERSHED TRANSFORM IN IMAGE SEGMENTATION

• ALGORITHM
• Image thresholding and denoising
• Distance and watershed transform use
• Extraction of individual segments
• Analysis of image components
• boundary signals and texture

11
10. FEATURE EXTRACTION AND CLASSIFICATION
RADON TRANSFORM IN ROTATION INVARIANT TEXTURE
FEATURES ESTIMATION

• ALOGORITHM
• Radon transform use for conversion of
• rotation to translation
• Translation invariant
• wavelet transform
• use for feature
• estimation
• Classification by
• neural networks

12
11. FEATURE BASED SEGMENTATION
FEATURE BASED BIOMEDICAL IMAGE SEGMENTATION

• PRINCIPLE
• Each root pixel of the original image is
  associated with its feature
• derived from its neighbourhood
• Pixels are individually classified into
  selected number of levels

13
12. CONCLUSION
COLLABORATION

• European Association for Signal and Image
  Processing
• IEE London, IEEE
• University of Cambridge, Brunel University, UK
• University Las Palmas, Spain

SELECTED PAPERS

• A. Procházka, I. Šindelárová, and J. Ptácek.
  Image De-noising and
• Restoration using Wavelet Transform . In
  European Control Conference
• ECC 2003 Conference Papers, Cambridge, UK,
  2003.
• A. Procházka and J. Ptácek. Wavelet Transform
  Application in
• Biomedical Image Recovery and Enhancement .
  In P. of 8th Multi-Conf.
• Systemics, Cybernetics and Informatic,
  Orlando, USA, 2004
• A. Procházka, A. Gavlasova, M. Mudrova. Rotation
  Invariant
• Biomedical Object Recognition. In Proc. of
  the EUSIPCO Conf.,
• EURASIP, Italy, 2006

14
Institute of Chemical Technology in
Prague Research Group of Digital Signal and Image
Processing
http // dsp.vscht.cz