Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.stsci.edu/stsci/meetings/irw/proceedings/perryk.dir/section3_7.html
Дата изменения: Sat Apr 16 04:20:11 1994
Дата индексирования: Sun Dec 23 19:34:34 2007
Кодировка:

Поисковые слова: annular solar eclipse
Experiments



Next: Conclusions Up: Generalized Cross-Validation as a Previous: Using the RGCV

Experiments

Two experiments were performed to demonstrate the RCGV stopping rule. The first was done with artificial data, and the second with HST data.

For the first experiment, an image was created that simulates a scene that might be typical for the wide-field planetary camera (see Fig. 1).

The image was then blurred using various ``simple'' blurs, and Poisson noise was added using an ADU gain of 8.0 e-/DN. Read-out noise was neglected to simplify the problem. The R-L algorithm was performed, and the RGCV stopping rule was computed. The mean-square error between the estimate and the original was also computed at each iteration. Finally, the -square measure was computed for further comparison. The results are summarized in Table 1.

The second group of experiments used the HST data for images of Jupiter, which are part of the simulation data available from the Space Telescope Science Institute database. The R-L algorithm was performed with the RGCV stopping rule. The results are shown in Fig. 2.


rlw@sundog.stsci.edu
Fri Apr 15 20:09:18 EDT 1994