Question
Comparison between curvelet and wavelets
What are the advantages in case of medical imaging and how are they different?
All Answers (7)
-
Wavelets capture space-frequency events such as linear edges, lines, texture, etc. I believe curvelets capture more circular type space-frequency events, such as arcs, etc. If you plan to localize linear lines, then wavelets are the way to go. but if the phenomenon you wish you capture has curved lines/ features, then curvelets would probably be better. -
Thanks April for giving the reply.
So according to u wavelets and curvelets can be applied on the same image. if the image contain linear lines and curves? -
I share the essential of April’s reply. In the same logic, one may similarly add Contourlet to the comparison(curves and irregular shapes):
JPEG 2000 that is the state-of-the-art image compression DICOM standard for storage and transmission of medical images- is wavelet based. Some researches in medical images compression showed limitations of wavelet transform as compared to contourlet transform at high compression ratios as in this paper (http://www.waset.org/journals/waset/v49/v49-148.pdf) for CT images.
For the classification use in CT, this paper also ends to the predictable supremacy of contourlet (http://www.ipcbee.com/vol16/12-E10005.pdf) while comparing wavelet, curvelet and contourlet approaches. -
Hello Puja - yes in fact you can apply both the waelets and curvelets to the same image - however the resultant wavelet ccoefficients will be different - one that localizes arc-like features whereas the other localizing lines like the edges of a square
-
Thank you Abdessamad Sir and April mam.
-
Can curvelets be used as a texture feature in images
-
Yes because texture is made of space frequency elements which are localized by the transform. But the direction and shape and curvature of the texture elements will influence which coefficients you get