Scribd
Upload
212 views

Image Processing in Biology11

The document discusses optimal edge-based shape detection from images. It introduces the concepts of edge detection, including types of edges and calculations for gradient magnitude and orientation. It then describes developing an optimal smoothing operator based on minimizing noise and error. The paper evaluates detecting edges and shapes for applications like vehicle detection. It finds the differential of Gaussian edge detector works best for localization, while the differential of oriented edge detector achieves better actual edge detection. Profiling shapes can then extend this to detect multiple shapes with high confidence.

Uploaded by

Waleed Abrar
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
212 views

Image Processing in Biology11

The document discusses optimal edge-based shape detection from images. It introduces the concepts of edge detection, including types of edges and calculations for gradient magnitude and orientation. It then describes developing an optimal smoothing operator based on minimizing noise and error. The paper evaluates detecting edges and shapes for applications like vehicle detection. It finds the differential of Gaussian edge detector works best for localization, while the differential of oriented edge detector achieves better actual edge detection. Profiling shapes can then extend this to detect multiple shapes with high confidence.

Uploaded by

Waleed Abrar
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 27

Background

Optimal Edge-Based Shape Detection


Univesity of Konstanz
By Waleed Abrar

February 5, 2015

Optimal Edge-Based Shape Detection

1 / 27

Background

Outline
1

Background

Introduction and Motivation

Optimal Edge Base Shape Detection

Experimental Results and Evaluations

Applications

Conclusion

Demo and References

Optimal Edge-Based Shape Detection

2 / 27

Background
Introduction and Motivation
Optimal Edge Base Shape Detection

Outline
1

Background

Introduction and Motivation

Optimal Edge Base Shape Detection

Experimental Results and Evaluations

Applications

Conclusion

Demo and References

Optimal Edge-Based Shape Detection

3 / 27

Background
Introduction and Motivation
Optimal Edge Base Shape Detection

Process of Image Acquisition

[1]

Optimal Edge-Based Shape Detection

4 / 27

Background
Introduction and Motivation
Optimal Edge Base Shape Detection

Sampling , Quantization and Level of Processing

Sampling
Digitization of the spatial coordinates (x,y)
Quantization
Digitization in amplitude (also called gray-level quantization)
Region Vs Boundary
Boundary sometime called as contour is a set of pixels in the
region that have one or more neighbour that are not in R

Optimal Edge-Based Shape Detection

5 / 27

Background
Introduction and Motivation
Optimal Edge Base Shape Detection

Outline
1

Background

Introduction and Motivation

Optimal Edge Base Shape Detection

Experimental Results and Evaluations

Applications

Conclusion

Demo and References

Optimal Edge-Based Shape Detection

6 / 27

Background
Introduction and Motivation
Optimal Edge Base Shape Detection

Low level processing

Figure: Lena

Optimal Edge-Based Shape Detection

7 / 27

Background
Introduction and Motivation
Optimal Edge Base Shape Detection

Type of Edges

[1]
Optimal Edge-Based Shape Detection

8 / 27

Background
Introduction and Motivation
Optimal Edge Base Shape Detection

Edge Calculations

Gradient Magnitude
Gradient Orientation
[1]

Optimal Edge-Based Shape Detection

9 / 27

Background
Introduction and Motivation
Optimal Edge Base Shape Detection

Edge Detection Problem

[1]
Optimal Edge-Based Shape Detection

10 / 27

Introduction and Motivation


Optimal Edge Base Shape Detection
Experimental Results and Evaluations

Outline
1

Background

Introduction and Motivation

Optimal Edge Base Shape Detection

Experimental Results and Evaluations

Applications

Conclusion

Demo and References

Optimal Edge-Based Shape Detection

11 / 27

Introduction and Motivation


Optimal Edge Base Shape Detection
Experimental Results and Evaluations

Main Concept in the Paper

Get better Detection of Edges + Better Localization of


Edge
Compare DOG, DOB and DODE
Minimizing the sum of the noise power and the mean
squared error between input and output

Optimal Edge-Based Shape Detection

12 / 27

Introduction and Motivation


Optimal Edge Base Shape Detection
Experimental Results and Evaluations

Calculations
Step edge is Corrupted with uniform white noise

We need to find h that minimises E2 + M 2


E = MSD between Input and output F
M = MSD of output Noise Responses

Optimal Edge-Based Shape Detection

13 / 27

Introduction and Motivation


Optimal Edge Base Shape Detection
Experimental Results and Evaluations

Edge Detection Cont..

[3]
Optimal Edge-Based Shape Detection

14 / 27

Introduction and Motivation


Optimal Edge Base Shape Detection
Experimental Results and Evaluations

Optimal Smoothing Operator


Wiener filter:

One Dimensional optimal filter

Two dimension can be extended as

Optimal Edge-Based Shape Detection

15 / 27

Optimal Edge Base Shape Detection


Experimental Results and Evaluations
Applications

Outline
1

Background

Introduction and Motivation

Optimal Edge Base Shape Detection

Experimental Results and Evaluations

Applications

Conclusion

Demo and References

Optimal Edge-Based Shape Detection

16 / 27

Optimal Edge Base Shape Detection


Experimental Results and Evaluations
Applications

Edge detection and vehicle detection

Optimal Edge-Based Shape Detection

17 / 27

Optimal Edge Base Shape Detection


Experimental Results and Evaluations
Applications

Edge Detection and Vehicle Detection

Optimal Edge-Based Shape Detection

18 / 27

Optimal Edge Base Shape Detection


Experimental Results and Evaluations
Applications

Concept of Profiling Shapes

Optimal Edge-Based Shape Detection

19 / 27

Optimal Edge Base Shape Detection


Experimental Results and Evaluations
Applications

Concept of Profiling Shapes..

Optimal Edge-Based Shape Detection

20 / 27

Experimental Results and Evaluations


Applications
Conclusion

Outline
1

Background

Introduction and Motivation

Optimal Edge Base Shape Detection

Experimental Results and Evaluations

Applications

Conclusion

Demo and References

Optimal Edge-Based Shape Detection

21 / 27

Experimental Results and Evaluations


Applications
Conclusion

Application Areas of optimal edge based shape detection

Optimal Edge-Based Shape Detection

22 / 27

Applications
Conclusion
Demo and References

Outline
1

Background

Introduction and Motivation

Optimal Edge Base Shape Detection

Experimental Results and Evaluations

Applications

Conclusion

Demo and References

Optimal Edge-Based Shape Detection

23 / 27

Applications
Conclusion
Demo and References

Summarizing paper
DOG vs DODE
DODE works really well for localization as compared to actual
detection of edge.
Profiling
Profiling Shapes extends the detection to multiple shapes with
high confidence
Enhancements
Combination of low level edge detection with mid level edge
grouping
Efficient
The Algorithm actually give better results with less computation.
Optimal Edge-Based Shape Detection

24 / 27

Applications
Conclusion
Demo and References

Recommendations

Recommendations
USE SSIM OR E-SSIM to achieve better results as well to
overcome orientation . SSIM give some kind of quanifiable
measure to further enhance the Algorithm.
- It also helps over come zooming scaled or rotated problem and
detection.
-MSE is not always the right choice because of the signal Fidelity.

Optimal Edge-Based Shape Detection

25 / 27

Applications
Conclusion
Demo and References

Outline
1

Background

Introduction and Motivation

Optimal Edge Base Shape Detection

Experimental Results and Evaluations

Applications

Conclusion

Demo and References

Optimal Edge-Based Shape Detection

26 / 27

Applications
Conclusion
Demo and References

Literature Review

[1]DigitalImageProcessing,Rafaelhttp:
//lit.fe.uni-lj.si/showpdf.php?lang=slo&type=doc&doc=dip&format=0
[2]http://www.slideshare.net/nchkarthik/digital-image-processing-26334694
[3]OptimalEdge-BasedShapeDetection..http:
//www.cfar.umd.edu/hankyu/shape_html/shape_html.html#fig:operator
PerformanceComparisonofMedianandWienerFilterinImageDe-noising,IJCA,2010volume12
ImageProcessingAndPatternRecognition(BITI3313)..http://de.mathworks.com/matlabcentral/
fileexchange/28757-tracking-red-color-objects-using-matlab

Optimal Edge-Based Shape Detection

27 / 27

You might also like