Hi,
To segment simultaneously a set of images (cosegmentation problem), i need to minimize an energy function which contain three terms : data term, smoothness term (like MRF model) and a global term that capture similarities between images. How can i minimize this energy function?
Thank you.
you can construct primarily a composite objective Energy function which includes the three terms (data term, smoothness and the global term). but you should be careful for the scaling process by either scaling the variables for the minimization or scaling the weights for each term.
Thank you for your reply Mr El-Rabaie,
I did not understand what do you mean by scaling the weights for each term can you explain me more please?
Thank you.
Scaling is a process should be carried out if the values of the variables you are looking to verify the minimization problem the gap between them is very large. The same concept is true if the objective function is composite every part of it should have a different weight. Unless you do that you will not be able to get a global minima.
Prof. S. EL-Rabaie
- Badges
- Science topic
- Similar topics
- Engineering
- Electrical Engineering
- Signal Processing
- Image Processing