Документ взят из кэша поисковой машины. Адрес оригинального документа : http://xmm.vilspa.esa.es/sas/5.4.1/doc/asmooth/node5.html
Дата изменения: Fri Jan 10 22:48:37 2003
Дата индексирования: Sat Dec 22 08:24:00 2007
Кодировка:

XMM-Newton Science Analysis System


asmooth (asmooth-2.10.2) [xmmsas_20030110_1802-5.4.1]

Meta Index / Home Page / Description / Adaptive smoothing

Method

In the adaptive mode, the value of $\sigma$ in equation 1 is made to vary (between bounds minsigma and maxsigma) in such a way that resulting the signa-to-noise ratio (SNR) is as close to uniform as possible. This requires that faint parts of the image, where the SNR starts out poorer, be smoothed more than bright parts. The SNR at a given pixel $p,q$ after smoothing is given by

$$snr_{p,q} = \frac{\sum_{i=-N}^{N} \sum_{j=-N}^{N} K_{i,j} \,... ...m_{j=-N}^{N} K_{i,j}^{2} \, \delta_{p+i,q+j}^{2}\right)^{1/2}}$$ (2)

where $y_{p,q}$ is the value of the unsmoothed image at that pixel and $\delta_{p,q}$ is the noise or uncertainty in that value. Because the variation in SNR with $\sigma$ is in general nonlinear, a simple procedure is employed in which, for each pixel, the task tests in turn each gaussian in the library until the SNR as calculated by equation 2 reaches the optimum value.