T. Frese, C. A. Bouman, N. C. Rouze, G. D. Hutchins, and K. Sauer. Bayesian multiresolution algorithm for PET reconstruction. In Proc. of IEEE Int'l Conf. on Image Proc., volume 2, pages 613--616, Vancouver, BC, Canada, September 10-13 2000.  Home/Search   Document Details and Download   Summary   Related Articles

....Similarly, by attaching a hidden Markov tree to m we can, in principle, synthesize the approach described here with that introduced in Example 6, a possibility that to date has not been examined in the literature. In addition, other nonlinear variations of this type of structure can be found in [125, 124] in which the estimation of wm based on xm takes on a form that is in some sense both more general and more restrictive than (46) In particular, in the models in these references, the individual detail coe#cients comprising wm are assumed to be conditionally independent when each is conditioned ....

....of the entire vector of coe#cients represent what in principle are restrictions compared to the general form of (46) although such restrictions are generally required in order to obtain computationally tractable models (e.g. we will see something similar in Section 6. 2) What is more general in [125, 124] is that the conditional distribution for each component of wm is modeled as Gaussian, with mean and variance that are nonlinear functions of the window of neighboring scaling coe#cients (see Section 6.3.3 for a further brief discussion) Note that both of the models (44) and (46) represent ....

[Article contains additional citation context not shown here]

T. Frese, C. A. Bouman, N. C. Rouze, G. D. Hutchins, and K. Sauer. Bayesian multiresolution algorithm for PET reconstruction. In Proc. of IEEE Int'l Conf. on Image Proc., volume 2, pages 613--616, Vancouver, BC, Canada, September 10-13 2000.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC