Directional multiscale modeling of images using the contourlet transform
read more
Citations
The contourlet transform: an efficient directional multiresolution image representation
The Nonsubsampled Contourlet Transform: Theory, Design, and Applications
Sparse directional image representations using the discrete shearlet transform
Optimally sparse multidimensional representation using shearlets
Sparse multidimensional representation using shearlets
References
Elements of information theory
A theory for multiresolution signal decomposition: the wavelet representation
A wavelet tour of signal processing
Density estimation for statistics and data analysis
The Laplacian Pyramid as a Compact Image Code
Related Papers (5)
The contourlet transform: an efficient directional multiresolution image representation
Frequently Asked Questions (10)
Q2. How can the HMT model be used to model the neighboring states?
In particular it has been shown that using only a mixture of two Gaussians, the HMT can already achieve satisfactory accuracy in wavelet coefficient modeling [7].
Q3. What is the reason for the horizontal distributions of the left sides of the plots?
If the authors assume that the horizontal distributions of the left sides of the plots are due to quantization errors and other sources of uncertainties dominating at small coefficient magnitudes, contourlet coefficients of natural images can be modeled according to some distributions with variances directly related to any linear combination of the magnitudes of their generalized neighborhoods.
Q4. What is the significance of joint statistics?
Joint statistics are particularly important because in the wavelet case, image processing algorithms exploiting joint statistics of coefficients [5]–[7],[11],[12] show significant improvements in performance over those that exploit marginal statistics alone [3],[8].
Q5. What is the significance of the mutual information estimation results for the three representative images?
Note that all images show significant mutual information across all of scale, space and directions, and reinforces their observation in Section III that coefficients are dependent on their generalized neighborhoods.
Q6. Why are the contourlet coefficients defined over different supports?
The reason is that the basis functions corresponding to the vertical and horizontal subbands are defined over different supports [19].
Q7. What is the widely used family of Gaussian mixture models?
The authors consider the hidden Markov model (HMM) family [7], which is one of the most well-known and widely used family of Gaussian mixture models.
Q8. What is the expected effect of contourlets on the directional filter?
Again note that as more effective directional filters for contourlets are developed in the future, it is expected that for contourlets, and should further decrease.
Q9. What is the definition of a directional filter bank?
The directional filter bank is a critically sampled filter bank that decomposes images into any power of two’s number of directions.
Q10. What is the value of contourlets in image processing?
Both results suggest that contourlets can capture directional information very well, which is a highly valuable property in image processing.