I processed a set of mammograpic images to improve the contrast and calculated five quality measures based on regions of interest. The improvement is based on wavelet transform.
Experimentation was performed with a set of wavelets bases and all combinations of decomposition levels to be processed, i.e. all subsets of decomposition levels. For example, I process the combination {1,2,3}, but also {1,2,4}, etc. The size of computation depends on the maximum level of decomposition of the wavelet base.
The question is: how to choose the best wavelet base and the best combination of decomposition levels to process according to the improvement measures used?
Dear Damian Valdés Santiago,
Look the links, may be useful.
https://www.eecis.udel.edu/~amer/CISC651/IEEEwavelet.pdf
https://www6.cityu.edu.hk/seam/web%20content/Papers/A%20novel%20technique%20for%20selecting%20mother%20wavelet%20function%20using%20an%20intelligent%20fault%20diagnosis%20system.pdf
Regards, Shafagat
Hi!
To choose the best wavelet base is needed to select experimentally this wavelet function by which you can obtain minimum levels for the decomposition.
You can obtain the best shrinkage decomposition on the base of some criterion, regard to the application: noise reduction, compression etc. For example one criterion is the minimum of entropy.
Best regards
Dears Shafagat Mahmudova and Veska Georgieva:
Thanks for response. The firstpart of my question is answered, but I don't know criteria for level combination selection. Do you know anyone?
Grateful, Damian
Hello
I can suggest you to look at this publications :
https://www.researchgate.net/publication/235933497_Noise_cancellation_in_Doppler_moment_estimation_of_array_weather_radar_signals_using_lifted_biorthogonal_wavelets
https://www.researchgate.net/publication/233966030_De-noised_estimation_of_theweather_Doppler_spectrum_by_the_wavelet_method
Cordially
Article Noise cancellation in Doppler moment estimation of array wea...
Article De-noised estimation of theweather Doppler spectrum by the w...
There are not general rules. It depends on application. The decomposition level must be small enough, to work with sub-bands containing a number of coefficients statistically significant. The decomposition level must be sufficiently big to obtain relevant wavelet coefficients magnitude values of the noiseless component of the input.
As for wavelets and their choices
I advise you to use several types of wavelet mothers with adaptive thresholding
- Badges
- Science topic
- Similar topics
- Engineering
- Electrical Engineering
- Signal Processing
- Image Processing