Scribd
Upload
29 views

An Enhanced Image Fusion Framework Using Morphological Operations Based Unsharp Masking

Dr. Sumanth Kumar Panguluri Research Work Presented in IEEE Conference

Uploaded by

Sumanth Kumar Panguluri
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
29 views

An Enhanced Image Fusion Framework Using Morphological Operations Based Unsharp Masking

Dr. Sumanth Kumar Panguluri Research Work Presented in IEEE Conference

Uploaded by

Sumanth Kumar Panguluri
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 18

An Enhanced Image Fusion Framework Using

Morphological Operations Based Unsharp Masking

Author 1 Author 2
P Sumanth Kumar Dr. Laavanya Mohan
Full Time Research Scholar Associate Professor
Department of ECE Department of ECE
VFSTR(deemed to be University) VFSTR(deemed to be University)

Department of
Electronics and Communication Engineering 1
Contents
1. Introduction
2. Objective of work
3. Block Diagram of Proposed Work
4. Experimental results
5. Conclusion

2
Introduction

a) Infrared Image b) Visible Image

c) Fused Image 3
Objective of work

 To construct an enhanced infrared and visible image fusion method for enhancing

visibility and improving situation awareness present in IR and VI images for

surveillance applications.

4
Block diagram of Proposed work

5
Procedure for morphological operations based unsharp
masking

6
Procedure for morphological operations based unsharp
masking

7
Morphology operations based unsharp masking output

Infrared image Visible image

Enhanced infrared image Enhanced visible image


8
Decomposition of curve-let transform

9
Formula for Fusion Rules
Fusion rule for Low-Frequency Coefficients
The formula for “PCA fusion rule” is given by

FPCA  P1 AI  P2 AV

Fusion rule for High-Frequency Coefficients

The formula for “Max fusion rule” is given by

FMax  max  DI , DV 

10
Fused results visual comparison for “Forest image”

a.) Infrared image b.) Visible image c.) Method-1 fused image d.) Method-2 fused image

e.) Method-3 fused image f.) Method-4 fused image g.) Proposed method fused image

11
TABLE 1. FUSED RESULTS METRICS VALUES COMPARISON
FOR “FOREST IMAGE”

Performance Metrics
Method
Standard Mean
Entropy Spatial frequency QAB/F
Deviation Gradient
Sharpen_filter 33.6649 6.7825 17.5139 24.4235 0.3312
PCA 32.8658 6.5140 19.7061 23.4752 0.1847
Curve-let 27.3125 6.5592 11.1400 21.9834 0.3123
Unsharp_masking 34.5629 6.7991 10.8214 24.2402 0.2864

Proposed 38.9024 7.0202 20.8464 25.0072 0.4125

12
Fused results visual comparison for “Sand-path image”

a.) Infrared image b.) Visible image c.) Method-1 fused image d.) Method-2 fused image

e.) Method-3 fused image f.) Method-4 fused image g.) Proposed method fused image

13
TABLE 2. FUSED RESULTS METRICS VALUES COMPARISON
FOR “SAND-PATH IMAGE”

Performance Metrics
Method
Standard Mean
Entropy Spatial frequency QAB/F
Deviation Gradient
Sharpen_filter 29.1907 6.7659 20.1871 35.0034 0.3308
PCA 27.7295 6.4784 22.3358 38.9181 0.1494
Curve-let 24.6842 6.5383 16.1316 27.8812 0.4236
Unsharp_masking 33.9778 6.9237 16.1049 31.2199 0.2533

Proposed 43.6718 7.2956 24.6780 42.0860 0.4358

14
Fused results visual comparison for “Garden image”

a.) Infrared image b.) Visible image c.) Method-1 fused image d.) Method-2 fused image

e.) Method-3 fused image f.) Method-4 fused image g.) Proposed method fused image

15
TABLE 3. FUSED RESULTS METRICS VALUES COMPARISON
FOR “GARDEN IMAGE”

Performance Metrics
Method
Standard Mean
Entropy Spatial frequency QAB/F
Deviation Gradient
Sharpen_filter 42.3256 7.0933 18.4584 39.2649 0.4170
PCA 38.2889 6.6142 18.1349 42.7031 0.1979
Curve-let 35.9385 6.8591 10.3522 24.1552 0.4165
Unsharp_masking 41.1372 6.9410 10.3008 26.3369 0.3275

Proposed 47.7884 7.2342 18.8402 45.2584 0.4280

16
Conclusion of Proposed work

 An enhanced IR and VI fusion method using morphological operations


based unsharp masking and curve-let transform is successfully constructed
in this paper for enhancing visibility and improving situation awareness.
 Three pairs of I-R and V-I images are used for testing the performance of
proposed method. Proposed method results are compared with four similar
existing fusion methods.
 Proposed method has produced much better results than latest similar
existing methods both in visual comparison and also in metric-value
comparison.

17
Thank you

18

You might also like