What is phase congruency principle?

In summary: From the link:Coherence means constant phase relationship.Phase congruency means they come together at some point, if only briefly.
  • #1
ramdas
79
0
Hello everyone,can anyone help me to understand the phase congruency principle. I have read that,in image processing if we reconstruct an image using its phase only,we can get obtain edges and lines and this is because of phase congruency principle.
I am searching more about phase congruence on web but i not understanding it properly.so can anyone explain what is phase congruency and its importance in image processing and other domain... thank u...
 
Engineering news on Phys.org
  • #2
Did you see this paper?

http://www.csse.uwa.edu.au/~pk/Research/research.html [Broken]

there's some attached MATLAB code that might provide the answers you're looking for.
 
Last edited by a moderator:
  • Like
Likes ramdas
  • #3
Hello everyone,can anyone help me to understand the phase congruency principle.
I am searching more about phase congruence on web but i not understanding it properly.so can anyone explain what is phase congruency ?Thank u...
 
  • #4
ramdas said:
... can anyone explain what is phase congruency ?
I thought jedishrfu's link was very informative on the subject. You do know what Edge detection is, right? From this link:
A recent development in edge detection techniques takes a frequency domain approach to finding edge locations. Phase congruency (also known as phase coherence) methods attempt to find locations in an image where all sinusoids in the frequency domain are in phase.
 
  • Like
Likes ramdas and jim hardy
  • #5
dlgoff said:
I thought jedishrfu's link was very informative on the subject. You do know what Edge detection is, right? From this link:
sir,sorry to disturb you.but as you said that "in phase congruency ,we find locations in an image where all sinusoids in the frequency domain are in phase." means as per your saying "edges and corners are in phase".Right! but sir,there can be other locations other than edges and corners which can be in phase . here,why "phase congruency concept doen't detect these locations which are in phase". So my question is why only edges and corners?why not other locations? If possible can you reply me using any simple image .Thank you.
 
  • #6
ramdas said:
means as per your saying "edges and corners are in phase".

edges and corners are where the Fourier components are maximally in phase
if you've ever manually (that is with pencil ruler and paper) plotted a square wave as sum of its first few Fourier terms it seems intuitive.
The steepest slopes occur where the Fourier terms come in phase, ie their individual steepest slopes (zero crossings) coincide.

Same author, more detailed presentation
http://www.csse.uwa.edu.au/~pk/research/pkpapers/phasecorners.pdf [Broken]
see section 2 fig1 on page 3, it shows that square wave exercise

upload_2015-4-18_8-21-27.png


were that part of an old fashioned black&white NTSC video signal, it'd change from very dark to very bright right there in the middle.
Note that's where all your components have briefly come into phase with one another. It's obvious from the picture but probably not from just the list of (nωt + Φ) terms
 
Last edited by a moderator:
  • Like
Likes ramdas
  • #7
jim hardy said:
edges and corners are where the Fourier components are maximally in phase
if you've ever manually (that is with pencil ruler and paper) plotted a square wave as sum of its first few Fourier terms it seems intuitive.
The steepest slopes occur where the Fourier terms come in phase, ie their individual steepest slopes (zero crossings) coincide.

Same author, more detailed presentation
http://www.csse.uwa.edu.au/~pk/research/pkpapers/phasecorners.pdf [Broken]
see section 2 fig1 on page 3, it shows that square wave exercise

View attachment 82189

were that part of an old fashioned black&white NTSC video signal, it'd change from very dark to very bright right there in the middle.
Note that's where all your components have briefly come into phase with one another. It's obvious from the picture but probably not from just the list of (nωt + Φ) terms
 
Last edited by a moderator:
  • #8
Thank u very much for your feedback.but whether phase congruency and phase coherence are same meaning?
dlgoff said:
I thought jedishrfu's link was very informative on the subject. You do know what Edge detection is, right? From this link:
 
  • #9
i don't think so.

Look up both definitions.

Coherence means constant phase relationship

congruence means they come together at some point, if only briefly.

from Webster's online

Full Definition of COHERENT

3
a : relating to or composed of waves having a constant difference in http://www.merriam-webster.com/dictionary/phase[1 [Broken] <coherent light>

Definition of CONGRUOUS
1
a : being in agreement, harmony, or correspondence

b : conforming to the circumstances or requirements of a situation : appropriate <a congruous room to work in — G. B. Shaw>
 
Last edited by a moderator:
  • Like
Likes ramdas

What is phase congruency principle?

The phase congruency principle is a concept in the field of image processing and computer vision that states that the phase of an image contains important information about its structure and features. It is based on the idea that images can be represented as a combination of amplitude and phase components, and that the phase component is a more reliable indicator of image structure than the amplitude component.

How is phase congruency principle different from other image processing techniques?

Unlike traditional image processing techniques that rely solely on the amplitude (brightness) of an image, the phase congruency principle takes into account the phase (position) information as well. This allows for a more accurate representation of image structure and features, especially in cases where images may contain similar amplitude values but different phase values.

What are some applications of the phase congruency principle?

The phase congruency principle has various applications in image processing and computer vision, including edge detection, image enhancement, feature extraction, and image registration. It is also commonly used in tasks such as motion analysis, texture analysis, and object recognition.

What are the limitations of the phase congruency principle?

While the phase congruency principle is a powerful concept, it also has some limitations. It may not work well in cases where images have low contrast or contain significant amounts of noise. Additionally, it may not be suitable for all types of images, such as those with very regular or symmetrical patterns.

How can the phase congruency principle be implemented in practice?

The phase congruency principle can be implemented using various techniques, such as Fourier transforms, Gabor filters, and steerable filters. These methods can be applied to images using software tools or programming languages such as MATLAB, Python, or OpenCV. There are also many research papers and tutorials available that provide step-by-step instructions on implementing the phase congruency principle.

Similar threads

Replies
13
Views
2K
Replies
7
Views
1K
  • Electrical Engineering
Replies
5
Views
5K
Replies
19
Views
2K
Replies
3
Views
445
  • Electrical Engineering
Replies
9
Views
921
  • Electrical Engineering
2
Replies
41
Views
3K
  • Electrical Engineering
Replies
12
Views
3K
Replies
49
Views
2K
Replies
1
Views
1K
Back
Top